CN117116432A - Disease characteristic processing method, device and equipment - Google Patents
Disease characteristic processing method, device and equipment Download PDFInfo
- Publication number
- CN117116432A CN117116432A CN202311370851.0A CN202311370851A CN117116432A CN 117116432 A CN117116432 A CN 117116432A CN 202311370851 A CN202311370851 A CN 202311370851A CN 117116432 A CN117116432 A CN 117116432A
- Authority
- CN
- China
- Prior art keywords
- matrix
- feature
- dimension reduction
- implicit
- explicit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 71
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 71
- 238000003672 processing method Methods 0.000 title abstract description 12
- 239000011159 matrix material Substances 0.000 claims abstract description 610
- 230000009467 reduction Effects 0.000 claims abstract description 244
- 238000012545 processing Methods 0.000 claims abstract description 135
- 238000009826 distribution Methods 0.000 claims abstract description 105
- 238000000034 method Methods 0.000 claims abstract description 48
- 230000009466 transformation Effects 0.000 claims description 26
- 238000010201 enrichment analysis Methods 0.000 claims description 22
- 238000011946 reduction process Methods 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 5
- 239000000523 sample Substances 0.000 description 121
- 230000008569 process Effects 0.000 description 23
- 238000004458 analytical method Methods 0.000 description 19
- 238000007689 inspection Methods 0.000 description 19
- 239000003814 drug Substances 0.000 description 18
- 238000004364 calculation method Methods 0.000 description 15
- 238000004422 calculation algorithm Methods 0.000 description 13
- 238000010586 diagram Methods 0.000 description 10
- 230000010354 integration Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 7
- 238000010801 machine learning Methods 0.000 description 7
- 238000012360 testing method Methods 0.000 description 7
- 238000000605 extraction Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000013507 mapping Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 102100029768 Histone-lysine N-methyltransferase SETD1A Human genes 0.000 description 2
- 101000865038 Homo sapiens Histone-lysine N-methyltransferase SETD1A Proteins 0.000 description 2
- 101150117538 Set2 gene Proteins 0.000 description 2
- 238000007418 data mining Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000556 factor analysis Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000003862 health status Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000000692 Student's t-test Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 238000000546 chi-square test Methods 0.000 description 1
- 239000013068 control sample Substances 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000002558 medical inspection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 210000001747 pupil Anatomy 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012353 t test Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Data Mining & Analysis (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Image Processing (AREA)
Abstract
The application discloses a disease feature processing method, device and equipment, which are characterized in that an initial feature matrix is generated through explicit features and medical sample data, and dimension reduction processing is carried out on the initial feature matrix to obtain a feature matrix after dimension reduction processing, wherein the feature matrix after dimension reduction processing comprises matrix elements corresponding to implicit features, and the implicit features are characterized by unknown features different from the explicit features; determining enrichment distribution parameters of case group data of each of the explicit features and the implicit features in the medical sample data based on the initial feature matrix and the feature matrix after the dimension reduction processing; and determining a target feature based on the enrichment distribution parameter. Based on the method, more unknown disease-related characteristics can be mined, so that the treatment of the disease characteristics is more comprehensive and reliable.
Description
Technical Field
The present application relates to the field of medical data mining, and more particularly, to a disease feature processing method, apparatus, and device.
Background
For a long time, the development of the inspection of the traditional Chinese medicine is seriously dependent on the naked eye observation and personal experience of doctors, and has great ambiguity, subjectivity and instability, which brings a plurality of invariants to the clinic, teaching, scientific research and the like of the inspection of the traditional Chinese medicine, and seriously limits the effect and popularization of the inspection of the traditional Chinese medicine, thus being particularly important to the scientific research and treatment of the characteristics of related diseases in the inspection process of the traditional Chinese medicine.
At present, research on disease characteristics in inspection of traditional Chinese medicine is focused on the relevance of patient characteristics and diseases, but current research objects are limited and lack of analysis on comprehensive characteristics.
Disclosure of Invention
In view of the above, the application provides a disease characteristic processing method, device and equipment, which are used for solving the problems of incomplete and inaccurate analysis of the existing traditional Chinese medicine inspection characteristics.
In order to achieve the above object, the following solutions have been proposed:
a method of treating a disease feature, comprising:
generating an initial feature matrix based on medical sample data, wherein each sample image in the medical sample data comprises at least one sub-region, determining a value of each element in the initial feature matrix based on whether an explicit feature exists in the sub-region of each of the sample images, the explicit feature characterizing a feature received at an input;
performing dimension reduction processing on the initial feature matrix to obtain a feature matrix subjected to dimension reduction processing, wherein the feature matrix subjected to dimension reduction processing comprises matrix elements corresponding to implicit features, and the implicit features represent unknown features different from the explicit features;
Performing enrichment analysis processing on the initial feature matrix and the feature matrix subjected to the dimension reduction processing, and determining enrichment distribution parameters of each explicit feature and each implicit feature in case group data of the medical sample data;
and determining a target feature based on the enrichment distribution parameter.
Optionally, the generating an initial feature matrix based on the medical sample data includes:
dividing each sample image into at least one sub-region, wherein the sub-region characterizes a region divided based on an observation object in the sample image;
generating a blank matrix based on the subareas of each sample image, wherein each blank matrix element in the blank matrix has a corresponding relation with a feature;
acquiring explicit characteristics input by a client;
respectively matching the explicit features with the features with corresponding relations of each blank matrix, and determining whether the features matched with the explicit features exist or not;
if so, marking the blank matrix element corresponding to the feature matched with the explicit feature as a first marking value;
if not, marking the blank matrix element corresponding to the feature which is not matched with the explicit feature as a second marking value;
An initial feature matrix is determined based on the blank matrix elements of the blank matrix that are labeled with the first label value and the second label value.
Optionally, when the initial feature matrix is a three-dimensional matrix of a region dimension, a feature dimension and a sample dimension, the performing the dimension reduction processing on the initial feature matrix to obtain a feature matrix after the dimension reduction processing includes:
performing dimension reduction processing on a matrix corresponding to the regional dimension of the initial feature matrix to obtain a regional dimension reduction matrix;
performing feature transformation on a matrix corresponding to the feature dimension of the initial feature matrix to obtain a feature transformation matrix;
dividing the matrix corresponding to the sample dimension of the initial feature matrix to obtain a case group matrix and a control group matrix;
performing dimension reduction treatment on the case group matrix and the control group matrix respectively to obtain a case group dimension reduction matrix and a control group dimension reduction matrix;
integrating the case group dimension reduction matrix with the control group dimension reduction matrix to obtain a sample dimension reduction matrix;
and integrating the regional dimension reduction matrix, the feature transformation matrix and the sample dimension reduction matrix to obtain a dimension reduction processed feature matrix corresponding to the initial feature matrix.
Optionally, the enriching analysis processing is performed on the initial feature matrix and the feature matrix after the dimension reduction processing, and determining enrichment distribution parameters of each of the explicit feature and the implicit feature in case group data of the medical sample data includes:
generating an initial reference matrix based on the initial feature matrix, wherein the initial reference matrix characterizes that all the explicit features exist in each sub-region of each sample image;
performing enrichment analysis processing on the initial feature matrix and the initial reference matrix, and determining a first enrichment distribution parameter of matrix elements corresponding to each explicit feature in the initial feature matrix in the initial reference matrix;
performing the dimension reduction processing on the initial reference matrix to obtain a dimension-reduced reference matrix;
carrying out enrichment analysis processing on the feature matrix after the dimension reduction processing and the reference matrix after the dimension reduction processing, and determining second enrichment distribution parameters of each matrix element corresponding to the implicit feature in the feature matrix after the dimension reduction processing;
an enrichment distribution parameter for each of the explicit feature and the implicit feature in case group data in the medical sample data is determined based on the first enrichment distribution parameter and the second enrichment distribution parameter.
Optionally, when more than one matrix exists in the area dimension of the feature matrix after the dimension reduction processing;
the enrichment analysis processing is performed on the feature matrix after the dimension reduction processing and the reference matrix after the dimension reduction processing, and a second enrichment distribution parameter of each matrix element corresponding to the implicit feature in the feature matrix after the dimension reduction processing in the reference matrix after the dimension reduction processing is determined, including:
acquiring a case group dimension reduction reference matrix and a control group dimension reduction reference matrix of the dimension reduction processed reference matrix in a sample dimension, and a case group dimension reduction matrix and a control group dimension reduction matrix of the dimension reduction processed feature matrix in the sample dimension;
selecting matrix data corresponding to the same group of samples from the case group dimension reduction reference matrix and the case group dimension reduction matrix to be combined to obtain a first subset;
selecting matrix data of the same group of samples corresponding to the first subset from the control group dimension reduction reference matrix and the control group dimension reduction matrix to be combined to obtain a second subset;
determining the frequency of occurrence index and the frequency of non-occurrence index of each matrix element corresponding to the implicit characteristic, which are respectively corresponding to the first subset and the second subset, wherein the frequency of occurrence index represents the frequency of occurrence of the matrix element corresponding to the implicit characteristic in the first subset or the second subset, and the frequency of non-occurrence index represents the frequency of non-occurrence of the matrix element corresponding to the implicit characteristic in the first subset or the second subset;
Determining whether each matrix element corresponding to the implicit characteristic in the case group dimension reduction reference matrix has the corresponding occurrence frequency index and the non-occurrence frequency index;
if so, determining a second enrichment distribution parameter of a reference matrix after the dimension reduction processing of each implicit feature based on the frequency indexes of occurrence and the frequency indexes of non-occurrence of each matrix element corresponding to the implicit feature in the first subset and the second subset respectively;
and if not, executing the steps of selecting matrix data corresponding to the same group of samples from the case group dimension reduction reference matrix and the case group dimension reduction matrix to be combined to obtain a first subset and then until each matrix element corresponding to the implicit feature in the case group dimension reduction reference matrix has the corresponding occurrence frequency index and the non-occurrence frequency index, and determining a second enrichment distribution parameter of the reference matrix corresponding to each implicit feature after dimension reduction processing based on the occurrence frequency index and the non-occurrence frequency index of each matrix element corresponding to the implicit feature in the first subset and the second subset respectively.
Optionally, the determining, based on the frequency of occurrence index and the frequency of non-occurrence index corresponding to each matrix element corresponding to the implicit feature in the first subset and the second subset, respectively, a second enrichment distribution parameter of a reference matrix after the dimension reduction processing for each implicit feature includes:
determining at least one case group observation based on the frequency of occurrence index and the frequency of non-occurrence index of each of the matrix elements corresponding to implicit features in the first subset;
determining at least one control group observation value based on the frequency of occurrence index and the frequency of non-occurrence index of each matrix element corresponding to an implicit feature in the second subset;
and based on the case group observed value and the control group observed value, checking, and taking the obtained significance result corresponding to each matrix element corresponding to the implicit characteristic as a second enrichment distribution parameter, wherein the significance result characterizes the distribution condition of the matrix element corresponding to the implicit characteristic in the case group sample of the medical sample data.
Optionally, the determining the target feature based on the enrichment distribution parameter includes:
Comparing the enrichment distribution parameters of each explicit feature or the matrix element corresponding to the implicit feature with a preset parameter range value to obtain comparison results corresponding to each explicit feature and the matrix element corresponding to the implicit feature;
and determining the explicit feature or the implicit feature of the matrix element corresponding to the explicit feature or the implicit feature of which the comparison result is a first comparison result as a target feature, wherein the first comparison result is that the enrichment distribution parameter is within a preset parameter range value.
Optionally, the method further comprises:
extracting feature parameters of the target features;
storing the characteristic parameters and the target characteristics in a database correspondingly;
and in response to receiving the feature information of the feature to be queried, matching the feature information with the feature parameters of each target feature in the database, and obtaining the target feature matched with the feature to be queried.
A disease signature processing apparatus comprising:
a matrix generation unit for generating an initial feature matrix based on medical sample data, wherein each sample image in the medical sample data comprises at least one sub-region, determining a value of each element in the initial feature matrix based on whether an explicit feature exists in the sub-region of each sample image, the explicit feature characterizing a received feature at an input;
The matrix dimension reduction unit is used for carrying out dimension reduction on the initial feature matrix to obtain a feature matrix after dimension reduction, wherein the feature matrix after dimension reduction comprises matrix elements corresponding to implicit features, and the implicit features represent unknown features different from the explicit features;
the parameter determining unit is used for carrying out enrichment analysis processing on the initial feature matrix and the feature matrix after the dimension reduction processing to determine enrichment distribution parameters of each explicit feature and each implicit feature in case group data of the medical sample data;
and the target determining unit is used for determining target characteristics based on the enrichment distribution parameters.
A disease signature processing apparatus comprising: a processor and a memory;
the memory is used for storing programs;
the processor is configured to execute the program to implement the steps of the method for treating a disease feature of any one of the above.
According to the technical scheme, the feature matrix with the matrix elements related to the implicit features is obtained by performing dimension reduction on the initial feature matrix, and more unknown disease related features, namely the implicit features, can be mined based on the feature matrix, so that the treatment of the disease features is more comprehensive.
The application also calculates the enrichment distribution parameters of the explicit feature and/or the implicit feature in the medical sample data, wherein the enrichment distribution parameters can reflect the distribution condition of the implicit feature in the medical sample data, and intuitively reflect the correlation between the explicit feature or the implicit feature and the disease in a numerical form, so that the finally obtained target feature is more accurate and reliable and has more reference significance.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present application, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a system architecture diagram of a method for implementing disease characteristics according to an embodiment of the present application;
FIG. 2 is a schematic flow chart of a treatment method for implementing disease features according to an embodiment of the present application;
FIG. 3 is a diagram illustrating an exemplary sample image according to an embodiment of the present application;
FIG. 4 is an exemplary diagram of an initial feature matrix provided by an embodiment of the present application;
FIG. 5 is an exemplary diagram of a feature matrix after dimension reduction processing according to an embodiment of the present application;
FIG. 6 is an exemplary diagram of an initial reference matrix and a reference matrix provided in an embodiment of the present application;
FIG. 7 is a schematic diagram of a disease feature processing apparatus according to an embodiment of the present application;
fig. 8 is a schematic structural diagram of a disease feature treatment apparatus according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The processing method of the disease characteristics can be used in the exploration process of clinical data of the inspection of the traditional Chinese medicine, or in the analysis process of the characteristics of the inspection of the disease by the traditional Chinese medicine, the relevant characteristics of the disease in the inspection process are explored, the known explicit characteristics and the implicit characteristics obtained by excavation are analyzed, and the target characteristics with strong relevance to the disease are determined.
In general, in the above process, the traditional Chinese medicine collects the characteristics that can be obviously observed in the inspection process corresponding to the case or the patient, that is, explicit characteristics, such as skin color, lip color, pupil color, wound and the like. The embodiment of the application processes the explicit feature to obtain the implicit feature which can correspond to the explicit feature, wherein the implicit feature is an unknown feature different from the explicit feature, and the data mining of the explicit feature can be completed based on the implicit feature. Further, the enrichment distribution parameters of the explicit features and the implicit features in the medical sample data are explored, the enrichment conditions of the explicit features and the implicit features are reflected, the implicit features or the explicit features with highest occurrence frequency can be determined, the target features are determined according to the correlation among the features, and the exploration of the relevant features of the diseases of the medical inspection is realized. In addition, the method can help doctors to improve the speed and accuracy of feature analysis in the process of consultation or research, and avoids subjectivity and ambiguity of feature analysis among different doctors.
In addition, as shown in fig. 1, the system architecture for implementing the disease feature processing method according to the embodiment of the present application may include a server 10 and a server 20, where the server 10 is configured to receive feature information corresponding to a patient input by a doctor, and the server 20 is configured to process feature information transmitted by the server 10 and return a processing result, that is, a target feature, to the server 10 for the doctor to view.
The server 20 may be a service device disposed on a network side of the server 10, and the server 10 may be a terminal device such as a computer or a mobile phone, and since a part of the terminal devices cannot store a large amount of medical sample data, the terminal devices are required to transmit the feature information to the server 20 to complete analysis processing of the feature information.
The feature information may be an explicit feature summarized by a doctor aiming at the health status of the patient after the doctor performs the inspection on the patient, and the doctor uploads the feature information to the server 10 to realize data exploration, analysis and processing of the feature information. The server 10 transmits the feature information to the server 20, and the server 20 extracts the implicit feature different from the feature information by converting the feature information into a matrix in digital form and performing dimension reduction processing on the matrix. Based on the analysis, the enrichment distribution conditions of the explicit features and the implicit features are analyzed, the enrichment distribution conditions can reflect the correlation between the explicit features or the implicit features and diseases, and based on the enrichment distribution conditions, the target features with strong correlation are extracted from a plurality of implicit features or the implicit features. The server 10 then presents the information to the doctor, who can then refer to the target features, feature information, and other medical experiences to more accurately analyze the health status of the patient.
Based on the above, the server 10 performs comprehensive data exploration according to the feature information uploaded by the doctor, and discovers unknown implicit features based on the explicit features, so that analysis on patients or cases is more comprehensive; in addition, the enrichment distribution condition based on the implicit characteristic and/or the explicit characteristic is combined, the correlation analysis between the disease and the characteristic is enhanced, and the finally obtained target characteristic can show comprehensiveness and accuracy.
Referring to fig. 2, a flow chart of a processing method for implementing the above disease features according to an embodiment of the present application is shown, where the flow chart may include:
step S110, generating an initial feature matrix based on the medical sample data.
Wherein each sample image in the medical sample data comprises at least one sub-region, the value of each element in the initial feature matrix being determined based on whether an explicit feature is present in the sub-region of each of the sample images, the explicit feature characterizing a received feature at an input.
The medical sample data is generated as samples based on a number of patient cases, wherein a number of sample images may be included, each image may be a matrix image, and the medical sample data may be a matrix data set. An exemplary diagram of a sample image is shown in fig. 3, which includes a plurality of blank matrix images. Each blank in the blank matrix image corresponds to an explicit feature, and each blank represents a blank matrix element, which is allowed to be assigned. Wherein R represents a set of regional dimension matrix images, each blank matrix image representing a sub-region; s represents a matrix image set of sample dimensions, it is understood that each column in the blank matrix image represents a sample S, and that the blank of a column corresponding to S1 is an explicit feature contained in sample S1, S1-S |s| Corresponding to s samples.
The subareas can be divided according to the observation objects of the traditional Chinese medicine inspection, if the traditional Chinese medicine inspection needs to observe the whole body or local spirit, color, shape, state and other aspects of a patient, the subareas can be divided according to the observed body parts of the patient, for example, the first subarea corresponds to eye features, the second subarea corresponds to tongue features, the features of the sample image can be divided into a plurality of subareas based on the characteristics, and the characteristics corresponding to all the subareas are combined to be the features of all the parts.
Based on the exemplary diagram shown in fig. 3, it is known that statistics of explicit features can be counted from both the region and the sample dimensions, and that it is also necessary to distinguish between explicit features of each input at the time of statistics, thereby re-recording the feature number. Based on the three dimensions of the region, the sample and the input explicit characteristics, the statistical results are obtained, and in the embodiment of the application, the statistical results are in a matrix form. The assignment of each element in the matrix characterizes the distribution of the explicit feature in the medical sample data. If the explicit features of an input are compared with the explicit features corresponding to the blank cells in the blank matrix image one by one, if the features are the same, the blank cells can be marked or assigned, and the explicit features of the input are confirmed to be distributed. Based on the method, the input explicit features can be subjected to feature comparison one by one, each blank lattice is assigned, and an initial feature matrix with corresponding values of each matrix element is obtained.
And step S120, performing dimension reduction processing on the initial feature matrix to obtain a feature matrix after the dimension reduction processing.
The feature matrix after the dimension reduction processing comprises matrix elements corresponding to implicit features, and the implicit features are characterized by unknown features different from the explicit features.
In the embodiment of the application, the dimension reduction processing of the initial feature matrix can be performed on the matrix corresponding to the three dimensions of the region, the sample and the feature respectively to obtain the matrix corresponding to the three dimensions after the dimension reduction processing, and then the three dimension reduction processed matrices are integrated to obtain the feature matrix corresponding to the initial feature matrix after the dimension reduction processing.
If the feature matrix after the dimension reduction processing can include matrix elements corresponding to and related to the implicit features, feature transformation is further required to be performed on the matrix carrying feature dimensions of the features in the initial feature matrix, so as to obtain matrix elements capable of representing the implicit features.
In addition, in the embodiment of the application, the dimension reduction processing mode for realizing dimension reduction on the initial feature matrix can be combined with the knowledge definition of the inspection of traditional Chinese medicine, an optimization algorithm, multi-dimension scale analysis, linear discriminant analysis and other processing methods, and the dimension reduction processing mode is not limited in a single way.
And step S130, carrying out enrichment analysis processing on the initial feature matrix and the feature matrix after the dimension reduction processing, and determining enrichment distribution parameters of each explicit feature and each implicit feature in case group data of the medical sample data.
Step S140, determining a target feature based on the enrichment distribution parameter.
In the embodiment of the application, the enrichment distribution parameters of each explicit feature in the initial feature matrix and each implicit feature in the feature matrix after the dimension reduction processing in the case group are required to be analyzed, wherein the case group is case group data generated based on case samples with specific diseases in the medical sample data, and the case group data comprises a plurality of explicit features related to the specific diseases. And the control group corresponding to the case group is control group data generated by case samples which do not suffer from the specific disease in the medical sample data, wherein the case samples which do not suffer from the specific disease can comprise health samples and case samples which suffer from other diseases but do not suffer from the specific disease.
In addition, in order to facilitate calculation of the enrichment distribution parameters of each implicit feature in the feature matrix after the dimension reduction processing in the case group data, the case group data may be converted into a matrix form and subjected to the dimension reduction processing for calculation.
The calculated enrichment distribution parameters of matrix elements corresponding to each implicit feature and/or explicit feature can determine the occurrence frequency of each implicit feature or explicit feature in the case group data. Assuming that the enrichment distribution parameter or the occurrence frequency corresponding to the matrix element for representing the implicit feature M meets a threshold, it can be determined that there is a more significant relationship between the implicit feature M and the feature contained in the case group data, and it can be further explained that there is a stronger correlation between the implicit feature M and the specific disease. Based on this, other implicit features similar to the implicit feature M are extracted as target features. The follow-up doctor can determine that when the target characteristics exist in the current diagnosed patient or case according to the output target characteristics, the target characteristics have stronger correlation with a certain disease, so that the doctor can judge the health state of the current patient or case by combining the traditional Chinese medicine inspection experience and other inspection results.
In summary, the embodiment of the application obtains the feature matrix with the matrix elements related to the implicit feature by performing the dimension reduction processing on the initial feature matrix, and based on the feature matrix, more unknown disease related features, namely the implicit features, can be mined, so that the treatment of the disease features by the application is more comprehensive.
In addition, the embodiment of the application also calculates the enrichment distribution parameters of the implicit characteristic and/or the explicit characteristic in the medical sample data, wherein the enrichment distribution parameters can reflect the occurrence frequency of the implicit characteristic or the explicit characteristic in the medical sample data, visually reflect the correlation between the implicit characteristic or the explicit characteristic and the disease in a numerical form, so that the finally obtained target characteristic is more accurate and reliable and has more reference significance.
Next, embodiments of the present application will be further described.
Based on the medical sample data and the entered explicit features, the process of determining the initial feature matrix may specifically include: dividing each sample image into at least one sub-region, wherein the sub-region characterizes a region divided based on an observation object in the sample image; generating a blank matrix based on the subareas of each sample image, wherein each blank matrix element in the blank matrix has a corresponding relation with a feature; acquiring explicit characteristics input by a client; respectively matching the explicit features with the features with corresponding relations of each blank matrix, and determining whether the features matched with the explicit features exist or not; if so, marking the blank matrix element corresponding to the feature matched with the explicit feature as a first marking value; if not, marking the blank matrix element corresponding to the feature which is not matched with the explicit feature as a second marking value; an initial feature matrix is determined based on the blank matrix elements of the blank matrix that are labeled with the first label value and the second label value.
Based on this, statistics can be performed on the sample images of the medical sample data for the explicit features received from the input, resulting in an initial feature matrix that characterizes the explicit features. Specifically, statistics may be performed based on the explicit features corresponding to each blank cell in the blank matrix image shown in fig. 3 and described above, whether the explicit features corresponding to each blank cell are the same as the explicit features received by the input end is determined one by one, and if the explicit features corresponding to a blank cells are the same as the explicit features received by one of the input end, the a blank cells may be marked as a first marking value to indicate that the a blank cells exist; if the explicit characteristics corresponding to the b blank cells are different from all the explicit characteristics received by the input end, the b blank cells can be marked as a second mark value to indicate that the b blank cells do not exist. The assignment of each element in the matrix is thus completed, resulting in an initial feature matrix that characterizes the distribution of the explicit features in the medical sample data.
In the embodiment of the application, statistics can be sequentially carried out on the sample images of the medical data samples according to the sequence of the explicit features received by the input end, so that the distribution condition of the explicit features in each region of the medical sample data can be obtained, and a multidimensional initial feature matrix is generated. Wherein the initial feature matrix may be referred to an exemplary plot of an initial feature matrix as shown in fig. 4. The matrix a of fig. 4 may be represented as where R, F, S represents a set of regions, features, samples, respectively, that refer to explicit features received at the input. In FIG. 4, F k Explicit feature representing kth input, S i Representing the ith sample, the matrix A contains a plurality of elements A r,i,k Wherein A is r,i,k The meaning of (c) can be determined with reference to the following formula (1):
(1)
in the above embodiment, the first flag value is 1, which indicates that the explicit feature of the input is distributed in this blank, and the second flag value is 0, which indicates that any explicit feature received at the input is not distributed at the matrix element marked 0.
Taking the matrix a as an example, when the matrix a is subjected to dimension reduction processing, the matrix a may be divided into matrices corresponding to regions, features and samples, and then the three matrices are subjected to dimension reduction respectively.
Performing dimension reduction processing on a matrix corresponding to the regional dimension of the initial feature matrix to obtain a regional dimension reduction matrix; performing feature transformation on a matrix corresponding to the feature dimension of the initial feature matrix to obtain a feature transformation matrix; dividing the matrix corresponding to the sample dimension of the initial feature matrix to obtain a case group matrix and a control group matrix; performing dimension reduction treatment on the case group matrix and the control group matrix respectively to obtain a case group dimension reduction matrix and a control group dimension reduction matrix; integrating the case group dimension reduction matrix with the control group dimension reduction matrix to obtain a sample dimension reduction matrix; and integrating the regional dimension reduction matrix, the feature transformation matrix and the sample dimension reduction matrix to obtain a dimension reduction processed feature matrix corresponding to the initial feature matrix.
The method comprises the steps of carrying out dimension reduction on a matrix of region dimensions, wherein the integration of sub-regions can be realized, and the dimension-reduced matrix of the region can only comprise one region. The dimension reduction processing mode can be used for carrying out region merging according to the knowledge definition of the inspection of the traditional Chinese medicine so as to realize region integration, and can also be used for carrying out dimension reduction processing on the matrix of the region dimension through the calculation methods such as linear weighting, polynomial fitting and the like determined by an optimization algorithm so as to realize region integration and obtain a region dimension reduction matrix.
The unknown implicit characteristics can be mined from the explicit characteristics by performing dimension reduction processing on the characteristic dimension matrix, the implicit characteristics are presented in a matrix form, and a characteristic transformation matrix is obtained, wherein matrix elements in the characteristic transformation matrix can represent the implicit characteristics. The feature transformation can be performed on the initial feature matrix through a machine learning algorithm, wherein the machine learning algorithm can adopt Principal Component Analysis (PCA), multidimensional scale analysis (MDS), factor Analysis (FA), linear Discriminant Analysis (LDA), recursive Feature Elimination (RFE), equal metric mapping (Isomap), sparse Regularization (SR), t-SNE, self-encoder (AE) and the like, and the feature transformation can be performed on the feature dimensional matrix based on the algorithm to obtain a feature transformation matrix. After feature transformation is performed on the matrix F corresponding to the feature dimension of the matrix A, the obtained feature transformation matrix is F ', wherein F' may be the matrix F itself or may be a combination of features in the matrix F, so that it is known that in the process, not only is integration and induction of explicit features completed, but also implicit features can be mined, and the obtained feature transformation matrix can comprise the explicit features and the implicit features, and is more comprehensive when enrichment distribution analysis is performed on the features later.
The dimension reduction processing is carried out on the matrix with the dimension of the sample, so that the effect of sample redundancy removal can be achieved, the matrix after dimension reduction is more simplified, and the subsequent calculation or analysis processing result for the matrix is more stable and has stronger generalization. When the dimension reduction treatment is carried out on the matrix of the sample dimension, the sample matrix can be divided into a case group matrix and a control group matrix according to whether a case has a specific disease or not. In this process, the dimension reduction mode adopted for the case group matrix and the control group matrix may adopt multiple or single downsampling or machine learning algorithm to reduce dimension, where if the dimension reduction is performed by adopting multiple downsampling, the dimension reduction results may be integrated by selecting modes such as average value and maximum value after multiple downsampling, so as to obtain the dimension reduction matrix. And if a machine learning algorithm is used, the dimension reduction can be performed by using any one of the algorithms listed above, but is not limited to the use of the machine learning algorithm.
And integrating the regional dimension reduction matrix, the feature transformation matrix and the sample dimension reduction matrix which are obtained after the dimension reduction processing of each dimension to obtain a feature matrix à which corresponds to the matrix A after the dimension reduction processing.
The embodiment of the application shown in fig. 5 provides a feature after dimension reductionReferring to FIG. 4, R ʹ, F ʹ, S ʹ in FIG. 5 represent the region, feature, and sample sets, respectively, and the feature matrix à after the dimension reduction of the matrix A shown can be described asThe mapping relationship between the matrix A and the matrix à before and after dimension reduction can be recorded as +.>。
In the matrix shown in fig. 5, the matrix elements are not blank as shown in the drawing, but each matrix element in the feature matrix obtained after the dimension reduction is very reduced according to different initial features does not have a fixed and unified rule, so in fig. 5, the blank matrix elements are taken as an example, and the matrix elements of the feature matrix after the dimension reduction are not specially limited. Further, for each explicit feature in the initial feature matrix a and each implicit feature in the reduced-dimension feature matrix Ã, the analysis process may include:
generating an initial reference matrix based on the initial feature matrix, wherein the initial reference matrix characterizes that all the explicit features exist in each sub-region of each sample image; performing enrichment analysis processing on the initial feature matrix and the initial reference matrix, and determining a first enrichment distribution parameter of matrix elements corresponding to each explicit feature in the initial feature matrix in the initial reference matrix; performing the dimension reduction processing on the initial reference matrix to obtain a dimension-reduced reference matrix; carrying out enrichment analysis processing on the feature matrix after the dimension reduction processing and the reference matrix after the dimension reduction processing, and determining a second enrichment distribution parameter of each matrix element related to the implicit feature in the feature matrix after the dimension reduction processing; an enrichment distribution parameter for each of the explicit feature and the implicit feature in case group data in the medical sample data is determined based on the first enrichment distribution parameter and the second enrichment distribution parameter.
Before calculating the enrichment distribution parameters of the implicit features in the dimensionality reduction case group, a reference matrix needs to be created, and the reference matrix can provide sample sums when calculating the enrichment distribution conditions. In the sample corresponding to the medical sample data, a plurality of cases exist in the case group and the control group respectively, but after the dimension reduction treatment, the matrix changes, and how many cases corresponding to the case group and the control group can not be correspondingly divided, so that an initial reference matrix needs to be re-created, and dimension reduction is performed to obtain the reference matrix.
The initial reference matrix and the initial feature matrix a belong to the same type of matrix, and the reference matrix and the feature matrix à after the dimension reduction process belong to the same type of matrix. An exemplary diagram of an initial reference matrix and a reference matrix is provided in accordance with an embodiment of the present application as shown in fig. 6. The initial reference matrix V is similar to the initial feature matrix A, but each matrix element in the initial reference matrix V is "1", and represents that each explicit feature exists, the matrix V can be marked as。
Then according to the mapping relation between the initial feature matrix A and the feature matrix à after the dimension reduction treatment, namelyPerforming dimension reduction on the initial reference matrix V to obtain a reference matrix +. >. The process of performing the dimension reduction processing on the initial reference matrix V may refer to the process of performing the dimension reduction processing on the initial feature matrix a to obtain the feature matrix Ã, where the operations performed on the initial reference matrix V and the initial feature matrix a in the dimension reduction processing are the same, and the dimension-reduced reference feature matrix is->The sample dimensions of (a) include a case group dimension reduction reference matrix and a control group dimension reduction reference matrix, which are not described in detail herein.
When enrichment analysis processing is carried out on matrix elements of the explicit feature or the implicit feature, analysis processing is carried out according to the corresponding relation among the matrices, for example, enrichment analysis processing is carried out on an initial feature matrix A and an initial reference matrix V, and a first enrichment distribution parameter related to the explicit feature is obtained; and (5) carrying out enrichment analysis treatment on the feature matrix à after the dimension reduction treatment and the reference matrix Ṽ after the dimension reduction treatment to obtain a second enrichment distribution parameter related to the implicit feature.
In the embodiment of the application, the enrichment analysis processing can be performed by using a mode of exploring the correlation between factors and diseases or a test mode of exploring the distribution difference of two groups of samples, and one mode can be selected to be realized at will.
In the embodiments of the present application, enrichment analysis concerning implicit features is described as an example. Matrix based on sample dimension of reference matrix after dimension reduction processingAnd the feature matrix à after the dimension reduction treatment calculates the enrichment distribution condition of each matrix element corresponding to the implicit feature in the case group dimension reduction reference matrix. In the embodiment of the application, the second enrichment distribution parameter of each implicit characteristic in the dimensionality reduction case group can be calculated by a method for exploring the related relation between factors and diseases, such as a calculation mode of chi-square test, OR value OR RR value and the like, and the occurrence frequency of each implicit characteristic in the case group is determined. It can be understood that the matrix V and the initial feature matrix a may also perform enrichment condition calculation according to the above manner, and determine the enrichment condition of each explicit feature in the initial feature matrix in the matrix V, so as to obtain a corresponding first enrichment condition parameter.
When the feature matrix à after the dimension reduction process has more than one region in the region dimension, the enrichment distribution parameter needs to be calculated for each sub-region in R' of the feature matrix à and each implicit feature on each sub-region in the above manner, and the specific calculation process may include:
Acquiring a case group dimension reduction reference matrix and a control group dimension reduction reference matrix of the dimension reduction processed reference matrix in a sample dimension, and a case group dimension reduction matrix and a control group dimension reduction matrix of the dimension reduction processed feature matrix in the sample dimension; selecting matrix data corresponding to the same group of samples from the case group dimension reduction reference matrix and the case group dimension reduction matrix to be combined to obtain a first subset; and selecting matrix data of the same group of samples corresponding to the first subset from the control group dimension reduction reference matrix and the control group dimension reduction matrix, and combining to obtain a second subset.
Determining the frequency of occurrence index and the frequency of non-occurrence index of each matrix element corresponding to the implicit characteristic, which are respectively corresponding to the first subset and the second subset, wherein the frequency of occurrence index represents the frequency of occurrence of the matrix element corresponding to the implicit characteristic in the first subset or the second subset, and the frequency of non-occurrence index represents the frequency of non-occurrence of the matrix element corresponding to the implicit characteristic in the first subset or the second subset; determining whether each matrix element corresponding to the implicit feature in the case group dimension reduction reference matrix has the corresponding frequency of occurrence index and the frequency of non-occurrence index.
If so, determining a second enrichment distribution parameter of a reference matrix after the dimension reduction processing of each implicit feature based on the frequency indexes of occurrence and the frequency indexes of non-occurrence of each matrix element corresponding to the implicit feature in the first subset and the second subset respectively.
And if not, executing the steps of selecting matrix data corresponding to the same group of samples from the case group dimension reduction reference matrix and the case group dimension reduction matrix to be combined to obtain a first subset and then until each matrix element corresponding to the implicit feature in the case group dimension reduction reference matrix has the corresponding occurrence frequency index and the non-occurrence frequency index, and determining a second enrichment distribution parameter of the reference matrix corresponding to each implicit feature after dimension reduction processing based on the occurrence frequency index and the non-occurrence frequency index of each matrix element corresponding to the implicit feature in the first subset and the second subset respectively.
The embodiment of the application needs to define two groups of samples, and respectively selects sample data representing the samples of the case group and the samples of the contrast group from a case group matrix and a contrast group matrix corresponding to sample dimensions of a reference matrix after dimension reduction processing and a feature matrix à after dimension reduction processing as a case group corresponding subset and a contrast group corresponding subset to determine a first subset corresponding to the samples of the case group And a second subset corresponding to the control group sample +.>。
Taking the first subset as an example, assuming that matrix data corresponding to three samples i, j and k in the case group dimension reduction reference matrix are selected to be used for combining the first subset, the matrix data selected from the case group dimension reduction matrix should also be matrix data corresponding to three samples i, j and k in the case group dimension reduction matrix, and the first subset is obtained based on the combination of the two groups of matrix data obtained above. The three samples i, j, k can be regarded as a group of samples, and optionally, the selected samples can be one or a group. Further, matrix data corresponding to samples i, j and k are selected from the dimension reduction reference matrix of the control group and the dimension reduction matrix of the control group respectively, and are combined to obtain a second subset.
Based on the four-way table corresponding to each feature in the calculated sample, the four-way table comprises an appearance index and a non-appearance index of each implicit feature in the case group sample and the control group sample respectively, the appearance index and the non-appearance index have the same meaning as the appearance frequency index and the non-appearance frequency main feature respectively, and the total of the appearance index and the non-appearance index can be calculated, wherein the appearance index and the non-appearance index can represent the feature In a specific case group sample or a specific control groupThe number of occurrences and the number of non-occurrences in the sample.
Specifically, the quadruple table can be understood with reference to table 1.
Table 1 quadruple table for each feature
Wherein,
。
further, the case group sample and the control group sample can be extracted from the sample dimensions of the reference matrix after the dimension reduction processing and the feature matrix à after the dimension reduction processing for a plurality of times to obtain a plurality of Set1 and Set2 combinations, and the four-way table corresponding to each implicit feature F ʹ in the feature matrix à after the dimension reduction processing is obtained as much as possible by performing the above calculation.
In addition, since the feature dimension matrix of the feature matrix à obtained after the dimension reduction processing may not only have unknown implicit features, but also may have integrated explicit features, in the embodiment of the present application, the enrichment distribution parameters of the explicit features existing in the feature matrix Ṽ may also be calculated, and further, the explicit features may also be determined as target features.
Further, based on the quadruple table corresponding to each feature, the chi-square value, chi-square check significance, OR value OR RR value and the like of the current feature are calculated by utilizing each index in the quadruple table, and one OR more numerical calculation results are obtained by the calculation and serve as enrichment distribution parameters of the current feature. When determining the target feature, comparing the enrichment distribution parameter of each explicit feature or the matrix element corresponding to the implicit feature with a preset parameter range value to obtain a comparison result corresponding to each explicit feature or the matrix element corresponding to the implicit feature; and determining the explicit feature or the implicit feature of the matrix element corresponding to the implicit feature or the explicit feature of which the comparison result is the first comparison result as a target feature.
Wherein, the process of calculating the enrichment distribution parameter for determining the current feature based on the indexes in the quadruple table may comprise: determining at least one case group observation based on the frequency of occurrence index and the frequency of non-occurrence index of each of the matrix elements corresponding to implicit features in the first subset; determining at least one control group observation value based on the frequency of occurrence index and the frequency of non-occurrence index of each matrix element corresponding to an implicit feature in the second subset; and based on the case group observed value and the control group observed value, checking, and taking the obtained significance result corresponding to each matrix element corresponding to the implicit characteristic as a second enrichment distribution parameter, wherein the significance result characterizes the distribution condition of the matrix element corresponding to the implicit characteristic in the case group sample of the medical sample data.
In the embodiment of the application, each column of data in table 1 is calculated according to a preset formula to obtain the corresponding observation value of each column. In table 1, if two columns of data exist, the data in the column of Set1 may be calculated based on a preset integration formula, and integrated into a value as an observed value of the case group. The data in the column of Set2 may be calculated based on a preset integration formula, and integrated into a value as an observation value of the comparison group, where the preset integration formula may be the following formula (2) or formula (3), and the calculation may be performed by one of the following formulas, which is not limited in any way.
(2)
(3)
In the foregoing scheme, a plurality of quadruplets can be obtained by sampling a plurality of times, and based on this, the comparison group observed value and the case group observed value corresponding to each quadruplet can be calculated as well. And further checking the observed values of the case group and the control group. The verification method may include, but is not limited to, a parameter mean test, a parameter variance test, a non-parameter mean test, a non-parameter variance test, and other test methods including a t-test, an Anova test, a rank sum test, etc., and based on the test methods, a significant result capable of characterizing the significance of the corresponding implicit feature distribution is obtained, whether the feature distribution in the feature matrix Ṽ is significant may be determined, and the significant result corresponding to each implicit feature is used as an enrichment distribution parameter.
The process of determining the target feature for the significance result based on the enrichment distribution parameter may include: comparing the enrichment distribution parameters of each explicit feature or the matrix element corresponding to the implicit feature with a preset parameter range value to obtain comparison results corresponding to each explicit feature and the matrix element corresponding to the implicit feature; and determining the explicit feature or the implicit feature of the matrix element corresponding to the explicit feature or the implicit feature of which the comparison result is a first comparison result as a target feature, wherein the first comparison result is that the enrichment distribution parameter is within a preset parameter range value.
According to the above, the enrichment distribution parameter may be a saliency result, and in the embodiment of the present application, the saliency result corresponding to each explicit feature or implicit feature may be compared with a preset saliency result threshold, and only the explicit feature or implicit feature corresponding to the saliency result satisfying the saliency result threshold condition may be determined as the target feature. And the preset significance result threshold may be determined based on known disease-related characteristics or physician case analysis experience.
In addition, the application may relate to case tracing of a patient, and the determined target feature and the feature parameter corresponding to the target feature may be stored in a preset database, so as to facilitate query of the patient or doctor. Optionally, the storing process may include: extracting feature parameters of the target features; storing the characteristic parameters and the target characteristics in a database correspondingly; and in response to receiving the feature information of the feature to be queried, matching the feature information with the feature parameters of each target feature in the database, and obtaining the target feature matched with the feature to be queried.
The feature parameters of the target feature may include information such as a case number and a patient name of the case, and store the feature parameters in a database. During query, feature information of the feature to be queried can be input, wherein the feature information can comprise part or all feature parameters of the target feature, or patient or case numbers, and the target feature matched with the feature to be queried is acquired by querying and matching corresponding information in a database based on the feature information.
In summary, the application explores the explicit and implicit characteristics of the disease diagnosed by the traditional Chinese medicine by combining the classical statistical method and the machine learning algorithm. The known explicit characteristics and the unknown implicit characteristics in the inspection of the traditional Chinese medicine are subjected to distributed statistical analysis, and the result is comprehensive, accurate and high in interpretation.
Based on this, the application method can be integrated and applied in the clinical inspection process of traditional Chinese medicine, specifically, referring to the description of fig. 1, a system architecture for implementing the disease feature processing method is formed based on the server side and the server, and further, the application of the system is described in detail below.
The above-described disease feature processing method is specifically applied to the server 20, in response to receiving the explicit feature transmitted from the server 10, the traditional Chinese medicine can also input a target disease at the server 10, i.e. analyze the target feature with strong correlation with the target disease from the explicit feature, so that the server 20 can divide the existing medical sample data into case combination comparison groups, so as to subsequently create an initial reference matrix for calculating the enrichment distribution parameters corresponding to the feature.
And generating an initial feature matrix according to the distribution condition of the explicit features on the medical sample data. Further adopting a linear weighting algorithm to the regional dimension matrix in the initial feature matrix to determine a regional dimension matrix; the feature dimension matrix is subjected to transformation of each feature in the feature dimension matrix by adopting an equal metric mapping algorithm, so that a feature transformation matrix is obtained; and performing dimension reduction on the case group matrix of the sample dimension by adopting a multi-time downsampling method, performing dimension reduction on the control group matrix by adopting a machine learning algorithm, and combining the two dimension reduction matrices to obtain the sample dimension reduction matrix. And combining the sample dimension reduction matrix, the region dimension reduction matrix and the feature transformation matrix to obtain a feature matrix corresponding to the initial feature matrix.
Further, case group samples and control group samples are extracted from sample dimensions of the feature matrix for multiple times, multiple groups of first subsets and second subsets are generated, OR values of each group of first subsets and second subsets are calculated, enrichment distribution parameters of features corresponding to each tetrad table are reflected, enrichment score parameters are screened based on a preset threshold value, the purpose of screening features is achieved, and finally target features with strong correlation with feature diseases are obtained.
The following describes a disease feature processing apparatus provided in an embodiment of the present application, and the disease feature processing apparatus described below and the disease feature processing method described above may be referred to correspondingly.
First, referring to fig. 7, a description will be given of a processing apparatus for a disease feature applied to the server 20, as shown in fig. 7, the processing apparatus for a disease feature may include:
a matrix generation unit 100, configured to generate an initial feature matrix based on medical sample data, where each sample image in the medical sample data includes at least one sub-region, and determine a value of each element in the initial feature matrix based on whether an explicit feature exists in the sub-region of each sample image, where the explicit feature characterizes a feature received by an input terminal;
the matrix dimension reduction unit 200 is configured to perform dimension reduction processing on the initial feature matrix to obtain a feature matrix after dimension reduction processing, where the feature matrix after dimension reduction processing includes matrix elements related to implicit features, and the implicit features characterize unknown features different from the explicit features;
a parameter determining unit 300, configured to perform enrichment analysis processing on the initial feature matrix and the feature matrix after the dimension reduction processing, and determine enrichment distribution parameters of each of the explicit feature and the implicit feature in case group data of the medical sample data;
The target determining unit 400 is configured to determine a target feature based on the enrichment distribution parameter.
According to the technical scheme, the feature matrix with the matrix elements related to the implicit features is obtained by performing dimension reduction on the initial feature matrix, and more unknown disease related features, namely the implicit features, can be mined based on the feature matrix, so that the treatment of the disease features is more comprehensive.
The embodiment of the application also calculates the enrichment distribution parameters of the explicit feature and/or the implicit feature in the medical sample data, wherein the enrichment distribution parameters can reflect the occurrence frequency of the implicit feature in the medical sample data, visually reflect the correlation between the explicit feature or the implicit feature and the disease in a numerical form, so that the finally obtained target feature is more accurate and reliable and has more reference significance.
Optionally, the matrix generating unit 100 includes:
a region dividing subunit configured to divide each of the sample images into at least one sub-region, the sub-region characterizing a region of the sample image divided based on an observation object;
a blank matrix generation subunit, configured to generate a blank matrix based on the sub-region of each sample image, where each blank matrix element in the blank matrix has a corresponding relationship with a feature;
The feature acquisition subunit is used for acquiring the explicit feature input by the client;
the feature matching subunit is used for respectively matching the explicit features with the features with corresponding relations of each blank matrix and determining whether the features matched with the explicit features exist or not;
an element marking first subunit, configured to mark, when a result of the feature matching subunit is yes, the blank matrix element corresponding to the feature that matches the explicit feature as a first marking value;
an element marking second subunit, configured to mark, when a result of the feature matching subunit is no, the blank matrix element corresponding to the feature that is not matched with the explicit feature as a second marking value;
a matrix acquisition subunit, configured to determine an initial feature matrix based on the blank matrix element marked with the first marking value and the second marking value in the blank matrix.
Optionally, when the initial feature matrix is a three-dimensional matrix of a region dimension, a feature dimension, and a sample dimension, the matrix dimension reduction unit 200 includes:
the regional dimension reduction subunit is used for carrying out dimension reduction processing on the matrix corresponding to the regional dimension of the initial feature matrix to obtain a regional dimension reduction matrix;
The characteristic transformation subunit is used for carrying out characteristic transformation on the matrix corresponding to the characteristic dimension of the initial characteristic matrix to obtain a characteristic transformation matrix;
the sample dividing subunit is used for dividing the matrix corresponding to the sample dimension of the initial feature matrix to obtain a case group matrix and a control group matrix;
the grouping dimension reduction subunit is used for respectively carrying out dimension reduction treatment on the case group matrix and the control group matrix to obtain a case group dimension reduction matrix and a control group dimension reduction matrix;
the grouping integration subunit is used for integrating the case group dimension reduction matrix with the control group dimension reduction matrix to obtain a sample dimension reduction matrix;
and the dimension matrix combination subunit is used for integrating the regional dimension reduction matrix, the feature transformation matrix and the sample dimension reduction matrix to obtain a dimension reduction processed feature matrix corresponding to the initial feature matrix.
Optionally, the parameter determining unit 300 includes:
an initial reference matrix generation subunit, configured to generate an initial reference matrix based on the initial feature matrix, where the initial reference matrix characterizes each of the sub-regions of each of the sample images that all of the explicit features exist;
The explicit parameter calculation subunit is used for carrying out enrichment analysis processing on the initial feature matrix and the initial reference matrix, and determining a first enrichment distribution parameter of matrix elements corresponding to each explicit feature in the initial feature matrix in the initial reference matrix;
the dimension reduction processing subunit is used for carrying out dimension reduction processing on the initial reference matrix to obtain a dimension-reduced reference matrix;
the implicit parameter calculation subunit is used for carrying out enrichment analysis processing on the feature matrix after the dimension reduction processing and the reference matrix after the dimension reduction processing, and determining a second enrichment distribution parameter of each matrix element corresponding to the implicit feature in the feature matrix after the dimension reduction processing;
an enrichment parameter determination subunit for determining an enrichment distribution parameter of case group data in the medical sample data for each of the explicit feature and the implicit feature based on the first enrichment distribution parameter and the second enrichment distribution parameter.
Optionally, when there is more than one matrix in the area dimension of the feature matrix after the dimension reduction processing, the parameter calculating subunit includes:
The matrix acquisition subunit is used for acquiring a case group dimension reduction reference matrix and a control group dimension reduction reference matrix of the dimension reduction processed reference matrix in a sample dimension, and a case group dimension reduction matrix and a control group dimension reduction matrix of the dimension reduction processed feature matrix in the sample dimension;
the case sample extraction subunit is used for selecting matrix data corresponding to the same group of samples from the case group dimension reduction reference matrix and the case group dimension reduction matrix to be combined to obtain a first subset;
a control sample extraction subunit, configured to select matrix data of the same set of samples corresponding to the first subset from the control set dimension reduction reference matrix and the control set dimension reduction matrix, and combine the matrix data to obtain a second subset;
a parameter determining subunit, configured to determine a frequency of occurrence index and a frequency of non-occurrence index of each matrix element corresponding to an implicit feature, where the frequency of occurrence index represents a frequency of occurrence of the matrix element corresponding to the implicit feature in the first subset or the second subset, and the frequency of non-occurrence index represents a frequency of non-occurrence of the matrix element corresponding to the implicit feature in the first subset or the second subset;
A feature parameter determining subunit, configured to determine whether each matrix element corresponding to an implicit feature in the case group dimension reduction reference matrix has the corresponding occurrence frequency index and the non-occurrence frequency index;
an enrichment parameter determining subunit, configured to determine, when a result of the feature parameter determining subunit is yes, a second enrichment distribution parameter of a reference matrix after the dimension reduction processing with respect to each implicit feature based on an occurrence frequency index and a non-occurrence frequency index corresponding to each matrix element corresponding to the implicit feature in the first subset and the second subset, respectively;
and the repeated extraction subunit is configured to execute, when the result of the feature parameter determining subunit is no, the step of the case sample extraction subunit until each matrix element corresponding to the implicit feature in the case group dimension reduction reference matrix has the corresponding frequency of occurrence index and the frequency of non-occurrence index, and determine, based on the frequency of occurrence index and the frequency of non-occurrence index of each matrix element corresponding to the implicit feature in the first subset and the second subset, respectively, a second enrichment distribution parameter of the reference matrix corresponding to each implicit feature after the dimension reduction process.
Optionally, the enrichment parameter determining subunit includes:
a case group observed value calculation unit configured to determine at least one case group observed value based on the occurrence frequency index and the non-occurrence frequency index of each matrix element corresponding to an implicit feature in the first subset;
a comparison group observed value calculation unit, configured to determine at least one comparison group observed value based on the occurrence frequency index and the non-occurrence frequency index of each matrix element corresponding to the implicit feature in the second subset;
and the observation value verification unit is used for verifying based on the case group observation value and the control group observation value, taking the obtained significance result corresponding to each matrix element corresponding to the implicit characteristic as a second enrichment distribution parameter, and the significance result represents the distribution condition of the matrix element corresponding to the implicit characteristic in the case group sample of the medical sample data.
Optionally, the target determining unit 400 includes:
and the threshold comparison subunit is used for comparing the enrichment distribution parameter of each explicit feature or the matrix element corresponding to the implicit feature with a preset parameter range value to obtain comparison results corresponding to each explicit feature and the matrix element corresponding to the implicit feature respectively.
And the target determining subunit is used for determining the explicit feature or the implicit feature of the matrix element corresponding to the explicit feature or the implicit feature of which the comparison result is the first comparison result as a target feature.
Optionally, the apparatus further comprises:
a parameter extraction unit, configured to extract feature parameters of the target feature;
a parameter storage unit, configured to store the feature parameter and the target feature in a database in correspondence;
and the query response unit is used for responding to the received characteristic information of the characteristic to be queried, matching the characteristic information with the characteristic parameters of each target characteristic in the database, and obtaining the target characteristic matched with the characteristic to be queried.
The disease characteristic processing device provided by the embodiment of the application can be applied to disease characteristic processing equipment.
Fig. 8 shows a schematic structural view of a disease-characteristic treatment apparatus, and referring to fig. 8, the structure of the disease-characteristic treatment apparatus may include: at least one processor 01, at least one memory 02, at least one communication bus 03 and at least one communication interface 04;
in the embodiment of the application, the number of the processor 01, the memory 02, the communication bus 03 and the communication interface 04 is at least one, and the processor 01, the memory 02 and the communication interface 04 complete the communication with each other through the communication bus 03;
Processor 01 may be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of the present invention, etc.;
the memory 02 may include a high-speed RAM memory, and may further include a non-volatile memory (non-volatile memory) or the like, such as at least one magnetic disk memory;
the memory stores a program, and the processor can call the program stored in the memory, wherein the program is used for realizing each processing flow in the processing scheme of the disease characteristics.
Finally, it is further noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of treating a disease feature, comprising:
generating an initial feature matrix based on medical sample data, wherein each sample image in the medical sample data comprises at least one sub-region, determining a value of each element in the initial feature matrix based on whether an explicit feature exists in the sub-region of each of the sample images, the explicit feature characterizing a feature received at an input;
Performing dimension reduction processing on the initial feature matrix to obtain a feature matrix subjected to dimension reduction processing, wherein the feature matrix subjected to dimension reduction processing comprises matrix elements corresponding to implicit features, and the implicit features represent unknown features different from the explicit features;
performing enrichment analysis processing on the initial feature matrix and the feature matrix subjected to the dimension reduction processing, and determining enrichment distribution parameters of each explicit feature and each implicit feature in case group data of the medical sample data;
and determining a target feature based on the enrichment distribution parameter.
2. The method of claim 1, wherein generating an initial feature matrix based on medical sample data comprises:
dividing each sample image into at least one sub-region, wherein the sub-region characterizes a region divided based on an observation object in the sample image;
generating a blank matrix based on the subareas of each sample image, wherein each blank matrix element in the blank matrix has a corresponding relation with a feature;
acquiring explicit characteristics input by a client;
respectively matching the explicit features with the features with corresponding relations of each blank matrix, and determining whether the features matched with the explicit features exist or not;
If so, marking the blank matrix element corresponding to the feature matched with the explicit feature as a first marking value;
if not, marking the blank matrix element corresponding to the feature which is not matched with the explicit feature as a second marking value;
an initial feature matrix is determined based on the blank matrix elements of the blank matrix that are labeled with the first label value and the second label value.
3. The method for processing disease features according to claim 1, wherein when the initial feature matrix is a three-dimensional matrix of a region dimension, a feature dimension, and a sample dimension, the performing a dimension reduction process on the initial feature matrix to obtain a feature matrix after the dimension reduction process includes:
performing dimension reduction processing on a matrix corresponding to the regional dimension of the initial feature matrix to obtain a regional dimension reduction matrix;
performing feature transformation on a matrix corresponding to the feature dimension of the initial feature matrix to obtain a feature transformation matrix;
dividing the matrix corresponding to the sample dimension of the initial feature matrix to obtain a case group matrix and a control group matrix;
performing dimension reduction treatment on the case group matrix and the control group matrix respectively to obtain a case group dimension reduction matrix and a control group dimension reduction matrix;
Integrating the case group dimension reduction matrix with the control group dimension reduction matrix to obtain a sample dimension reduction matrix;
and integrating the regional dimension reduction matrix, the feature transformation matrix and the sample dimension reduction matrix to obtain a dimension reduction processed feature matrix corresponding to the initial feature matrix.
4. The method for processing disease features according to claim 1, wherein the performing enrichment analysis processing on the initial feature matrix and the feature matrix after the dimension reduction processing to determine enrichment distribution parameters of each of the explicit feature and the implicit feature in case group data of the medical sample data includes:
generating an initial reference matrix based on the initial feature matrix, wherein the initial reference matrix characterizes that all the explicit features exist in each sub-region of each sample image;
performing enrichment analysis processing on the initial feature matrix and the initial reference matrix, and determining a first enrichment distribution parameter of matrix elements corresponding to each explicit feature in the initial feature matrix in the initial reference matrix;
performing the dimension reduction processing on the initial reference matrix to obtain a dimension-reduced reference matrix;
Carrying out enrichment analysis processing on the feature matrix after the dimension reduction processing and the reference matrix after the dimension reduction processing, and determining second enrichment distribution parameters of each matrix element corresponding to the implicit feature in the feature matrix after the dimension reduction processing;
an enrichment distribution parameter for each of the explicit feature and the implicit feature in case group data in the medical sample data is determined based on the first enrichment distribution parameter and the second enrichment distribution parameter.
5. The method according to claim 4, wherein when there is more than one matrix in the area dimension of the feature matrix after the dimension reduction process;
the enrichment analysis processing is performed on the feature matrix after the dimension reduction processing and the reference matrix after the dimension reduction processing, and a second enrichment distribution parameter of each matrix element corresponding to the implicit feature in the feature matrix after the dimension reduction processing in the reference matrix after the dimension reduction processing is determined, including:
acquiring a case group dimension reduction reference matrix and a control group dimension reduction reference matrix of the dimension reduction processed reference matrix in a sample dimension, and a case group dimension reduction matrix and a control group dimension reduction matrix of the dimension reduction processed feature matrix in the sample dimension;
Selecting matrix data corresponding to the same group of samples from the case group dimension reduction reference matrix and the case group dimension reduction matrix to be combined to obtain a first subset;
selecting matrix data of the same group of samples corresponding to the first subset from the control group dimension reduction reference matrix and the control group dimension reduction matrix to be combined to obtain a second subset;
determining the frequency of occurrence index and the frequency of non-occurrence index of each matrix element corresponding to the implicit characteristic, which are respectively corresponding to the first subset and the second subset, wherein the frequency of occurrence index represents the frequency of occurrence of the matrix element corresponding to the implicit characteristic in the first subset or the second subset, and the frequency of non-occurrence index represents the frequency of non-occurrence of the matrix element corresponding to the implicit characteristic in the first subset or the second subset;
determining whether each matrix element corresponding to the implicit characteristic in the case group dimension reduction reference matrix has the corresponding occurrence frequency index and the non-occurrence frequency index;
if so, determining a second enrichment distribution parameter of a reference matrix after the dimension reduction processing of each implicit feature based on the frequency indexes of occurrence and the frequency indexes of non-occurrence of each matrix element corresponding to the implicit feature in the first subset and the second subset respectively;
And if not, executing the steps of selecting matrix data corresponding to the same group of samples from the case group dimension reduction reference matrix and the case group dimension reduction matrix to be combined to obtain a first subset and then until each matrix element corresponding to the implicit feature in the case group dimension reduction reference matrix has the corresponding occurrence frequency index and the non-occurrence frequency index, and determining a second enrichment distribution parameter of the reference matrix corresponding to each implicit feature after dimension reduction processing based on the occurrence frequency index and the non-occurrence frequency index of each matrix element corresponding to the implicit feature in the first subset and the second subset respectively.
6. The method according to claim 5, wherein determining the second enriched distribution parameter of the reference matrix after the dimension reduction processing with respect to each of the implicit features based on the frequency of occurrence index and the frequency of non-occurrence index of each of the matrix elements corresponding to the implicit features in the first subset and the second subset, respectively, comprises:
determining at least one case group observation based on the frequency of occurrence index and the frequency of non-occurrence index of each of the matrix elements corresponding to implicit features in the first subset;
Determining at least one control group observation value based on the frequency of occurrence index and the frequency of non-occurrence index of each matrix element corresponding to an implicit feature in the second subset;
and based on the case group observed value and the control group observed value, checking, and taking the obtained significance result corresponding to each matrix element corresponding to the implicit characteristic as a second enrichment distribution parameter, wherein the significance result characterizes the distribution condition of the matrix element corresponding to the implicit characteristic in the case group sample of the medical sample data.
7. The method of claim 1, wherein determining a target feature based on the enriched distribution parameter comprises:
comparing the enrichment distribution parameters of each explicit feature or the matrix element corresponding to the implicit feature with a preset parameter range value to obtain comparison results corresponding to each explicit feature and the matrix element corresponding to the implicit feature;
and determining the explicit feature or the implicit feature of the matrix element corresponding to the explicit feature or the implicit feature of which the comparison result is a first comparison result as a target feature, wherein the first comparison result is that the enrichment distribution parameter is within a preset parameter range value.
8. The method of treating a disease feature of claim 1, further comprising:
extracting feature parameters of the target features;
storing the characteristic parameters and the target characteristics in a database correspondingly;
and in response to receiving the feature information of the feature to be queried, matching the feature information with the feature parameters of each target feature in the database, and obtaining the target feature matched with the feature to be queried.
9. A disease characterized treatment device comprising:
a matrix generation unit for generating an initial feature matrix based on medical sample data, wherein each sample image in the medical sample data comprises at least one sub-region, determining a value of each element in the initial feature matrix based on whether an explicit feature exists in the sub-region of each sample image, the explicit feature characterizing a received feature at an input;
the matrix dimension reduction unit is used for carrying out dimension reduction on the initial feature matrix to obtain a feature matrix after dimension reduction, wherein the feature matrix after dimension reduction comprises matrix elements corresponding to implicit features, and the implicit features represent unknown features different from the explicit features;
The parameter determining unit is used for carrying out enrichment analysis processing on the initial feature matrix and the feature matrix after the dimension reduction processing to determine enrichment distribution parameters of each explicit feature and each implicit feature in case group data of the medical sample data;
and the target determining unit is used for determining target characteristics based on the enrichment distribution parameters.
10. A disease characterized treatment apparatus comprising: a processor and a memory;
the memory is used for storing programs;
the processor being adapted to execute the program to carry out the steps of the method of treatment of disease features according to any one of claims 1-8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311370851.0A CN117116432B (en) | 2023-10-23 | 2023-10-23 | Disease characteristic processing device and equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311370851.0A CN117116432B (en) | 2023-10-23 | 2023-10-23 | Disease characteristic processing device and equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117116432A true CN117116432A (en) | 2023-11-24 |
CN117116432B CN117116432B (en) | 2023-12-15 |
Family
ID=88809480
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311370851.0A Active CN117116432B (en) | 2023-10-23 | 2023-10-23 | Disease characteristic processing device and equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117116432B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108231201A (en) * | 2018-01-25 | 2018-06-29 | 华中科技大学 | A kind of construction method, system and the application of disease data analyzing and processing model |
CN109119133A (en) * | 2018-08-03 | 2019-01-01 | 厦门大学 | Tcm clinical practice big data storage method based on multiple labeling feature selecting and classification |
CN110598722A (en) * | 2018-06-12 | 2019-12-20 | 清华大学 | Multi-modal neuroimaging data automatic information fusion system |
CN112750531A (en) * | 2021-01-21 | 2021-05-04 | 广东工业大学 | Automatic inspection system, method, equipment and medium for traditional Chinese medicine |
WO2022056013A1 (en) * | 2020-09-08 | 2022-03-17 | Kang Zhang | Artificial intelligence for detecting a medical condition using facial images |
KR20230019393A (en) * | 2021-07-30 | 2023-02-08 | 고려대학교 산학협력단 | Individual pathogen simulator model in brain using reaction-diffusion equation on intrinsic connectivity |
WO2023025255A1 (en) * | 2021-08-27 | 2023-03-02 | 之江实验室 | Multi-center medical diagnosis knowledge graph representation learning method and system |
KR20230106205A (en) * | 2022-01-05 | 2023-07-13 | 동아대학교 산학협력단 | Apparatus and methods for classifying neurodegenerative diseases image of amyloid-positive based on deep-learning |
-
2023
- 2023-10-23 CN CN202311370851.0A patent/CN117116432B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108231201A (en) * | 2018-01-25 | 2018-06-29 | 华中科技大学 | A kind of construction method, system and the application of disease data analyzing and processing model |
CN110598722A (en) * | 2018-06-12 | 2019-12-20 | 清华大学 | Multi-modal neuroimaging data automatic information fusion system |
CN109119133A (en) * | 2018-08-03 | 2019-01-01 | 厦门大学 | Tcm clinical practice big data storage method based on multiple labeling feature selecting and classification |
WO2022056013A1 (en) * | 2020-09-08 | 2022-03-17 | Kang Zhang | Artificial intelligence for detecting a medical condition using facial images |
CN112750531A (en) * | 2021-01-21 | 2021-05-04 | 广东工业大学 | Automatic inspection system, method, equipment and medium for traditional Chinese medicine |
KR20230019393A (en) * | 2021-07-30 | 2023-02-08 | 고려대학교 산학협력단 | Individual pathogen simulator model in brain using reaction-diffusion equation on intrinsic connectivity |
WO2023025255A1 (en) * | 2021-08-27 | 2023-03-02 | 之江实验室 | Multi-center medical diagnosis knowledge graph representation learning method and system |
KR20230106205A (en) * | 2022-01-05 | 2023-07-13 | 동아대학교 산학협력단 | Apparatus and methods for classifying neurodegenerative diseases image of amyloid-positive based on deep-learning |
Non-Patent Citations (2)
Title |
---|
蔡艺芳;李灿东;: "面部望诊二维坐标体系的建立", 福建中医学院学报, no. 01 * |
郭睿;王忆勤;颜建军;李福凤;燕海霞;许朝霞;: "中医舌诊的客观化研究", 中国中西医结合杂志, no. 07 * |
Also Published As
Publication number | Publication date |
---|---|
CN117116432B (en) | 2023-12-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112329964B (en) | Method, device, equipment and storage medium for pushing information | |
US20170147753A1 (en) | Method for searching for similar case of multi-dimensional health data and apparatus for the same | |
CN110299209B (en) | Similar medical record searching method, device and equipment and readable storage medium | |
CN110931128B (en) | Method, system and device for automatically identifying unsupervised symptoms of unstructured medical texts | |
CN111445968A (en) | Electronic medical record query method and device, computer equipment and storage medium | |
CN110782998A (en) | Data auditing method and device, computer equipment and storage medium | |
CN111104481B (en) | Method, device and equipment for identifying matching field | |
CN114358001A (en) | Method for standardizing diagnosis result, and related device, equipment and storage medium thereof | |
WO2021008601A1 (en) | Method for testing medical data | |
CN111415760B (en) | Doctor recommendation method, doctor recommendation system, computer equipment and storage medium | |
CN110910991A (en) | Medical automatic image processing system | |
CN111833984A (en) | Medicine quality control analysis method, device, equipment and medium based on machine learning | |
CN111243753A (en) | Medical data-oriented multi-factor correlation interactive analysis method | |
CN107193996B (en) | Similar medical record matching and retrieving system | |
CN118312816A (en) | Cluster weighted clustering integrated medical data processing method and system based on member selection | |
CN113436725B (en) | Data processing method, system, computer device and computer readable storage medium | |
CN113779295A (en) | Retrieval method, device, equipment and medium for abnormal cell image features | |
WO2021047376A1 (en) | Data processing method, data processing apparatus and related devices | |
CN117116432B (en) | Disease characteristic processing device and equipment | |
CN113010783A (en) | Medical recommendation method, system and medium based on multi-modal cardiovascular disease information | |
CN114168751B (en) | Medical text label identification method and system based on medical knowledge conceptual diagram | |
CN110010231A (en) | A kind of data processing system and computer readable storage medium | |
CN115700826A (en) | Receipt processing method, receipt display method, receipt processing device, receipt display device, computer equipment and storage medium | |
CN105989094B (en) | Image retrieval method based on middle layer expression of hidden layer semantics | |
CN113095432A (en) | Visualization system and method based on interpretable random forest |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |