CN111883247A - Analysis system for correlation between behavior data and medical outcome - Google Patents
Analysis system for correlation between behavior data and medical outcome Download PDFInfo
- Publication number
- CN111883247A CN111883247A CN202010742790.6A CN202010742790A CN111883247A CN 111883247 A CN111883247 A CN 111883247A CN 202010742790 A CN202010742790 A CN 202010742790A CN 111883247 A CN111883247 A CN 111883247A
- Authority
- CN
- China
- Prior art keywords
- medical
- data
- behavior
- patient
- module
- 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
Images
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
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
- G16H40/20—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Theoretical Computer Science (AREA)
- General Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Optimization (AREA)
- Mathematical Physics (AREA)
- Operations Research (AREA)
- Pure & Applied Mathematics (AREA)
- Strategic Management (AREA)
- Mathematical Analysis (AREA)
- Biomedical Technology (AREA)
- Epidemiology (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Primary Health Care (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Databases & Information Systems (AREA)
- Computational Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Pathology (AREA)
- Evolutionary Biology (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Probability & Statistics with Applications (AREA)
- Life Sciences & Earth Sciences (AREA)
- Algebra (AREA)
- Quality & Reliability (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
Abstract
The invention discloses an analysis system for correlation between behavior data and medical outcome, which comprises an operation processor, and a data acquisition unit, a data analysis module, a database and a result output module which are connected with the operation processor, wherein the data acquisition unit acquires the operation behavior of medical personnel and the external sign information of a patient so as to acquire the behavior data; the data acquisition unit acquires diagnostic data of a patient; the data analysis module analyzes the acquired data; the database stores the acquired and analyzed data; and the result output module outputs the analysis result. According to the medical behavior data acquisition and analysis system and method, the behavior data of the medical staff and the physical signs and behavior data of the patient are acquired and analyzed to judge the influence of the behavior data on the medical result, so that the medical behavior establishment is facilitated in a standardized and targeted manner, the recovery time of the patient is shortened, and the medical service quality is improved.
Description
Technical Field
The invention belongs to the technical field of medical treatment, and particularly relates to an analysis system for correlation between behavior data and medical outcome.
Background
In the development of intelligent hospital construction, the improvement of the quality and efficiency of medical services by using intelligent and informatization technologies is a direction with urgent needs and serious challenges, and various intelligent terminal devices and medical auxiliary systems are produced at the same time. The existing products and technologies are mainly based on data such as electronic medical records, paper documents, medical images and the like, and provide quick information query and diagnosis and treatment assistance for medical workers through quick information retrieval and classification and data analysis on certain single diseases.
However, the medical outcome of a general patient is influenced by various factors, and in addition to the real-time dynamic optimization of the medical scheme, the final medical outcome is changed by the medical behavior of the medical care personnel and the corresponding fine behaviors such as the dietary habits, the work and rest rules and the like of the patient during the whole diagnosis and treatment process. Scientific intervention to the misbehavior of patients undergoing diagnosis and treatment becomes a major challenge in current medical development.
Disclosure of Invention
The invention aims to: in order to solve the problem that the quality and efficiency of medical services in the existing intelligent information medical service development process cannot be guaranteed, an analysis system for correlation between behavior data and medical outcomes is provided.
The technical scheme adopted by the invention is as follows:
an analysis system for correlation of behavior data and medical outcome comprises an operation processor, a data acquisition unit, a data analysis module, a database and a result output module, wherein the data acquisition unit, the data analysis module, the database and the result output module are connected with the operation processor; the data acquisition unit acquires diagnostic data of a patient; the data analysis module analyzes the acquired data; the database stores the acquired and analyzed data; the result output module outputs the analysis result, the operation processor is also connected with a display unit, a reminding unit and a standard correction module, and the display unit displays the acquired and acquired data and displays the analysis result; the reminding unit is used for reminding the analyzed abnormal data; the standard correction module corrects the medical staff to carry out the irregular error behaviors in the medical behaviors and the behaviors influencing the final medical result.
Wherein, data acquisition unit includes and nurses monitor probe and wear-type camera, and the monitor probe of doctorsing and nurses uses ball formula monitoring machine.
Wherein the data unit comprises a structuring preprocessing module, a feature representation module and a correlation analysis module.
The processing flow of the structured preprocessing module comprises the following steps: firstly, converting diagnosis and treatment records into vectors with time sequence information according to structured text data corresponding to medical behavior judgment indexes and medical operation standard standards, wherein each element in the vectors is a judgment value of each operation flow according to the judgment indexes or a logical value of whether to operate compliance;
comparing the acquired time sequence label of the video behavior with the judgment index by taking the patient identity as an index, and storing a logic value of whether to respond to the compliance operation to a corresponding position of the operation;
then, carrying out standardization processing on the formed vector to obtain a medical action standardization vector;
generating a patient state response vector corresponding to each element of the medical behavior standardization vector, wherein the vector stores the response state of the patient under the types of eating habits, work and rest rules and motion modes when the corresponding medical behavior is finished, and the corresponding element is a logic value for judging whether to respond or not;
adding the indexed data into the acquired time sequence label representing the state of the patient;
and finally, forming a large sparse matrix by the patient state response vectors according to a time sequence, wherein each row of the matrix represents a time sequence vector corresponding to the patient behavior.
The structured preprocessing module converts diagnosis and treatment records into time sequence information according to operation flows according to structured text data corresponding to medical behavior judgment indexes and medical operation standard standards, grades and quantifies each operation flow according to the judgment indexes, standardizes formed vectors, and converts the diagnosis and treatment records in the electronic medical record into a medical behavior standardized vector M1×LCorresponding to element mtT is more than or equal to 1 and less than or equal to L, and is a judgment score between 0 and 1 or a logical value 0/1 of whether to operate the compliance;
comparing the acquired time sequence label of the video behavior with the judgment index, and storing the logic value whether to respond to the compliance operation to the corresponding position of the medical behavior standardization vector, corresponding to the element mt0/1; in this module, the corresponding medical action standardization vector is each element mtA patient state response vector is generatedThe vector stores the response state of the patient in the categories of eating habits, work and rest rules, motion patterns and the like when the medical behavior is completed at the moment t, and corresponding elementsthe ith behavior category at the time t, i is more than or equal to 1 and less than or equal to N, the logical value 0/1 of response is obtained, the patient state response vectors form a sparse matrix according to time sequence, and the patient behaviors of the ith category form a patient behavior time sequence vectorElements at time t
The characteristic representation module adopts a DBM (database management Module) architecture model to cooperatively represent the characteristic representation of various types of behavior data, including various types of patient states and medical behavior states, each RBM architecture model stacked in the characteristic representation module is used for learning the time sequence representation of one type of behavior data, and the module performs characteristic extraction in the hierarchical learning process, divides the characteristic into two outputs and expands weight vectors of various types of behaviors and the retained behavior data of main characteristics.
Wherein, the representation flow of the characteristic representation module is as follows: the DBM architecture model consists of a visible layer and t hidden layers and is used as the stack of t RBM architecture models;
the feature weight vector finally generated by the moduleRepresenting the distribution of various behavior data at the Kth time node for the characteristic representation of various behavior categories obtained by learning the behavior data at K moments, and performing iterative training by taking K as a time step to obtain the dynamic characteristic representation of various behavior data;
the behavior data of the main features is composed of F matrixes, wherein F is the number of reserved main features and the matrix And representing the behavior data of all samples corresponding to the jth reserved characteristic at K moments.
The correlation analysis module uses an antagonistic neural network joint probability distribution estimation method for analysis, and comprises two models, wherein the first model takes the medical effect of the current time node as a label, the trained model predicts the medical effect of the next time node, the prediction result is compared with the corresponding actual medical effect, and when the model effect is reduced, the model is subjected to iterative optimization to obtain reasonable dynamic prediction on the medical outcome; and the model II takes the final medical outcome as a label, the trained model predicts the final medical outcome of the new sample, and the analysis data is obtained by combining the feature weight vector output by the last module.
Wherein the correlation analysis module comprises two models, a model jointGAN1 for predicting the medical effect at the time of the next time node T +1, and an input vector YnextEach element of (a) indicates whether the medical outcome corresponding to the behavior data at the K moments is improved, if the improvement is stored as 1, otherwise, the improvement is 0, and the first prediction result of the model is yT+1If the actual result Y at the time of T +1 is observedT+1And yT+1If the model is not equal to the preset value, the model I is trained again; model two jointGAN2 predicts the final medical outcome, with the input vector YfinalEach element of (a) corresponds to the final medical outcome of the sample, with the predicted result being yfinal。
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. according to the medical behavior data acquisition and analysis system and method, the behavior data of the medical staff and the physical signs and behavior data of the patient are acquired and analyzed to judge the influence of the behavior data on the medical result, so that the medical behavior establishment is facilitated in a standardized and targeted manner, the recovery time of the patient is shortened, and the medical service quality is improved.
2. According to the invention, the current medical effect is considered by combining the current behavior data of medical staff and patients, and the method of the antagonistic neural network joint probability distribution estimation is utilized to respectively act the medical effect of the current time predicted at the last time and the currently generated medical result on the model, so that the final medical outcome is reasonably predicted, and the reference is provided for the real-time analysis and judgment of the medical effect by the medical staff.
Drawings
FIG. 1 is a system block diagram of the present invention;
FIG. 2 is a system block diagram of a data acquisition unit according to the present invention;
FIG. 3 is a system block diagram of a data analysis module of the present invention;
FIG. 4 is a system block diagram of a structured pre-processing module of the present invention;
FIG. 5 is a system block diagram of a feature representation module of the present invention;
FIG. 6 is a system block diagram of a correlation analysis module according to the present invention.
The labels in the figure are: 1. an arithmetic processor; 10. a data acquisition unit; 20. a data acquisition unit; 30. a data analysis module; 40. a database; 50. a display unit; 60. a result output module; 70. a reminding unit; 80. standardizing a correction module; 101. a medical care monitoring probe; 102. a head-mounted camera; 301. a structured pre-processing module; 302. a feature representation module; 303. and a correlation analysis module.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1 to 3, an analysis system for correlation between behavior data and medical outcome includes an operation processor 1, where the operation processor 1 may be a locally deployed computer or a cloud-based processor providing computation, and a data acquisition unit 10, a data acquisition unit 20, a data analysis module 30, a database 40, and a result output module 60 connected to the operation processor 1, where the data acquisition unit 10 is configured to acquire operation behaviors of medical staff and external sign information of a patient, so as to acquire behavior data, and unlike a conventional method that simply depends on pathological characteristics of a diseased part of the patient to implement corresponding medical measures, in this embodiment, a reference data amount serving as medical information of the patient is further increased, so as to expand medical analysis correlation, improve accuracy of medical means selection, and thus perform more scientific treatment, the treatment effect is improved on the basis of the traditional treatment, the treatment rehabilitation period of the patient is shortened, the problem of shortage of hospital hospitalization resources is relieved, and the treatment expense of the patient is reduced; the data acquisition unit 20 is configured to acquire diagnostic data of a patient, and unlike the data acquisition unit 10, the acquired data is medical data generated in a series of procedures of inquiry, diagnosis, hospitalization, and treatment of the patient, including but not limited to a diagnostic book, a nursing record, and clinical signs of the patient, and needs to be accessed to a medical system of an existing hospital; the data analysis module 30 is used for analyzing the acquired data, analyzing whether medical behaviors of medical staff for patients are standardized and behavior degree in the diagnosis and treatment process, thereby providing accurate intervention, improving the quality of medical services, reasonably predicting final medical outcome according to the behavior data of the medical staff and the patients and combining the current medical effect, and providing reference for real-time analysis and judgment of the medical staff on the medical effect; the database 40 is used for storing the acquired and analyzed data and providing data support for the normal operation of the system; the result output module 60 is used for outputting the analysis result, so that the medical staff can check the analysis result more clearly and directly;
the operation processor 1 is also connected with a display unit 50, a reminding unit 70 and a standard correction module 80, wherein the display unit 50 is a display screen and is used for displaying acquired and acquired data and displaying an analysis result, so that the data is convenient to watch and the experience of medical staff is expanded; the reminding unit 70 is used for reminding the analyzed abnormal data, if a certain physical sign data of a patient is abnormal, and a certain operation of a medical worker is wrong, the data which can generate a large or serious influence on a final medical result is taken as a data reminder and confirmed again by a worker, and is also used for reminding directly and obviously, the standard correcting module 80 is used for correcting irregular wrong behaviors in medical behaviors of the medical worker and behaviors influencing the final medical result, so that specific treatment can be fully realized according to specific conditions of different people, and the final medical result is ensured;
the data acquisition unit 10 comprises a medical care monitoring probe 101 and a head-mounted camera 102, the medical care monitoring probe 101 is an existing monitoring probe of a hospital, meanwhile, in order to ensure that a patient and medical care behaviors can be effectively acquired, a ball-type monitoring machine is recommended to be used, and the head-mounted camera 102 obtains the change of a shooting visual field by following the head movement of a medical worker, so that the patient sign behavior data and the medical behavior data of the medical worker can be more clearly obtained;
the data unit 30 includes a structured preprocessing module 301, a feature representation module 302, and a correlation analysis module 303, where the structured preprocessing module 301 is configured to perform data structured processing on acquired and acquired data and store the data in a database, and a specific process is to convert a diagnosis and treatment record into a vector with time sequence information according to a medical action judgment index and structured text data corresponding to a medical operation specification standard, where each element in the vector is a judgment score or a logical value of whether to operate a compliance rule for each operation flow according to the judgment index;
then, for diagnosis and treatment operations which have requirements in the operation flow but cannot be provided in the electronic medical record, the patient identity is taken as an index, the time sequence label of the collected video behavior is compared with the judgment index, and whether a logic value responding to the compliance operation is stored to the corresponding position of the operation or not is stored;
then, the formed vector is subjected to standardization treatment, so that a medical action standardization vector is obtained;
generating a patient state response vector corresponding to each element of the medical behavior standardization vector, wherein the vector stores the response state of the patient under the categories of eating habits, work and rest rules, motion modes and the like when the corresponding medical behavior is finished, and the corresponding element is a logic value for judging whether to respond or not;
the information which can not be extracted from the diagnosis and treatment records is similar to the operation in the second step, but the acquired time sequence label representing the state of the patient is added into the indexed data;
finally, forming a large sparse matrix by the patient state response vectors according to a time sequence, wherein each row represents a time sequence vector corresponding to one patient behavior;
taking data of 1 patient in 1 diagnosis and treatment process as an example, as shown in fig. 4:
according to the structured text data corresponding to the medical behavior judgment index and the medical operation standard, the diagnosis and treatment records are converted into the structured text data with time sequence information according to the operation flows, each operation flow is graded and quantized according to the judgment index, and the formed vector is subjected to standardization treatment, so that the diagnosis and treatment records in the electronic medical record are converted into a medical behavior standardization vector M1×LCorresponding to element mt(t is more than or equal to 1 and less than or equal to L for the medical operation corresponding to the time t) is a judgment value between 0 and 1 or a logical value 0/1 for judging whether to operate the compliance;
for diagnosis and treatment operations which are required in the operation flow but cannot be provided in the electronic medical record, the patient identity is taken as an index, the time sequence label of the collected video behavior is compared with the judgment index, and whether the logical value of the compliance operation is responded or not is stored to the corresponding position of the medical behavior standardization vector, and the corresponding element m ist0/1;
many times the patient's behavioral state is uncontrolled and has no specific criteria for evaluation, in this module, every element m of the corresponding medical behavior standardized vectortA patient state response vector is generatedThe vector stores the response state of the patient in the categories of eating habits, work and rest rules, motion patterns and the like when the medical behavior is completed at the moment t, and corresponding elements(i-th behavior category at time t, i is more than or equal to 1 and less than or equal to N) is a logic value 0/1 of response, the operation in the second step cannot be executed by the information extracted from the structured data, the matrix formed by the patient state response vector according to the time sequence is a sparse matrix, and the patient behaviors of the i-th category form a patient behavior time sequence vectorElements at time t
In this embodiment, the feature representation module 302 cooperatively represents the feature representation of various types of behavior data (including various types of patient states and medical behavior states) by using a DBM architecture model, each RBM architecture model stacked therein is used for learning the time sequence representation of one type of behavior data, and the module automatically performs feature extraction (each feature represents 1 type of behavior) in a hierarchical learning process;
the module will eventually have two outputs:
the weight vector comprises various behavior categories, which reflect the contribution rate of data corresponding to each characteristic (various behaviors) in the behavior data of the medical procedure, namely, the importance degree of each behavior is visually seen so as to guide medical staff to pertinently declare a patient before the medical procedure is started;
only the behavior data of the main characteristics are reserved so as to improve the training efficiency of the next module analysis model;
taking the extraction of behavior data in K time instants as an example, assuming that the number of samples (number of patients) is num, as shown in fig. 5, a DBM architecture model is composed of a visible layer and t hidden layers, and is regarded as a stack of t RBM architecture models, and in particular, we put label information on the top layer;
the feature weight vector finally generated by the moduleThe behavior data at K times are learned to obtain characteristic representation of various behavior categories. It visually represents the distribution of various types of behavior data at the Kth time node. In addition, performing iterative training by taking K as a time step, thereby obtaining dynamic characteristic representation of various behavior data;
the behavior data for which only the main features are reserved consists of F matrices, F being the number of main features reserved, the matrix And representing the behavior data of all samples corresponding to the jth reserved characteristic at K moments.
In this embodiment, the correlation analysis module 303 predicts the medical effect of each step based on the joint probability distribution corresponding to all the features by using a method of estimating the joint probability distribution of the antagonistic neural network and using the behavior data extracted by the feature of the previous module. The module establishes two models simultaneously to perform two tasks simultaneously;
the first model takes the medical effect of the current time node as a label, and the trained model can predict the medical effect of the next time node. The prediction result is compared with the corresponding actual medical effect, and when the effect of the model is reduced, the model is subjected to iterative optimization, so that reasonable dynamic prediction on the medical outcome is obtained, and reference is provided for the medical staff to analyze and judge the medical effect in real time;
the model II takes the final medical outcome as a label, the trained model is used for predicting the final medical outcome of a new sample (a new patient), and reference data are provided for medical staff and medical researchers by combining the feature weight vector output by the last module;
as shown in FIG. 6, the purpose of model-jointGAN 1 is to predict the effect of the treatment at the next time node T +1, with an input vector YnextEach element of (a) indicates whether the medical outcome corresponding to the behavior data at the K moments is improved, if the improvement is stored as 1, otherwise, the improvement is 0, and the first prediction result of the model is yT+1If the actual result Y at the time of T +1 is observedT+1And yT+1If the two labels are not equal, the first model is trained again by using the new label;
the purpose of model bijointGAN 2 is to predict the final medical outcome, with the input vector YfinalEach element of (a) corresponds to the final medical outcome of the sample, with the predicted result being yfinal。
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (9)
1. The analysis system for the correlation between the behavior data and the medical outcome is characterized by comprising an operation processor (1), and a data acquisition unit (10), a data acquisition unit (20), a data analysis module (30), a database (40) and a result output module (60) which are connected with the operation processor (1), wherein the data acquisition unit (10) acquires the operation behavior of medical staff and the external sign information of a patient so as to acquire the behavior data; a data acquisition unit (20) acquires diagnostic data of a patient; the data analysis module (30) analyzes the acquired data; the database (40) stores the acquired and analyzed data; the result output module (60) outputs the analysis result, the operation processor (1) is also connected with a display unit (50), a reminding unit (70) and a standard correction module (80), and the display unit (50) displays the acquired and acquired data and displays the analysis result; a reminding unit (70) for reminding the analyzed abnormal data; the normative correction module (80) corrects irregular error behaviors in medical actions of medical staff and behaviors influencing final medical results.
2. The system for analyzing dependency of behavioral data on medical outcome of claim 1, wherein the data acquisition unit (10) comprises a health care monitoring probe (101) and a head-mounted camera (102), the health care monitoring probe (101) using a ball monitor.
3. A system for analysis of relevance of behavioural data to a medical outcome according to claim 1, characterised in that the data unit (30) comprises a structured preprocessing module (301), a feature representation module (302) and a relevance analysis module (303).
4. A system for analysis of dependency of behavioural data on medical outcome as claimed in claim 3, wherein the processing flow of the structured pre-processing module (301) comprises the following: (1) firstly, converting diagnosis and treatment records into vectors with time sequence information according to structured text data corresponding to medical behavior judgment indexes and medical operation standard standards, wherein each element in the vectors is a judgment value of each operation flow according to the judgment indexes or a logical value of whether to operate compliance;
(2) comparing the acquired time sequence label of the video behavior with the judgment index by taking the patient identity as an index, and storing a logic value of whether to respond to the compliance operation to a corresponding position of the operation;
(3) then, carrying out standardization processing on the formed vector to obtain a medical action standardization vector;
(4) then generating a response vector of the patient state corresponding to each element of the medical behavior standardization vector, wherein the vector stores the response state of the patient under the types of eating habits, work and rest rules and motion modes when the corresponding medical behavior is finished, and the corresponding element is a logic value for judging whether to respond or not;
(5) adding the indexed data into the acquired time sequence label representing the state of the patient;
(6) and finally, forming a large sparse matrix by the patient state response vectors according to a time sequence, wherein each row of the matrix represents a time sequence vector corresponding to the patient behavior.
5. The system for analyzing the correlation between the behavioral data and the medical outcome of claim 4, wherein the structured preprocessing module (301) converts the medical records into time series information according to the operational procedures based on the structured text data corresponding to the medical behavior evaluation index and the medical operation standard, classifies and quantifies each operational procedure according to the evaluation index, standardizes the formed vectors, and converts the medical records in the electronic medical record into a standardized vector M for the medical behavior1×LCorresponding to element mtT is more than or equal to 1 and less than or equal to L, and is a judgment score between 0 and 1 or a logical value 0/1 of whether to operate the compliance;
comparing the acquired time sequence label of the video behavior with the judgment index, and storing the logic value whether to respond to the compliance operation to the corresponding position of the medical behavior standardization vector, corresponding to the element mt0/1; in this module, the corresponding medical action standardization vector is each element mtA patient state response vector is generatedThe vector stores the response state of the patient in the categories of eating habits, work and rest rules, motion patterns and the like when the medical behavior is completed at the moment t, and corresponding elementsthe ith behavior category at the time t, i is more than or equal to 1 and less than or equal to N, the logical value 0/1 of response is obtained, the patient state response vectors form a sparse matrix according to time sequence, and the patient behaviors of the ith category form a patient behavior time sequence vectorElements at time t
6. The system for analyzing correlations of behavioral data and medical outcomes of claim 3, wherein the feature representation module (302) employs DBM architectural models to cooperatively represent the features of various categories of behavioral data, including various categories of patient states and medical behavioral states, each RBM architectural model stacked therein for learning a time-series representation of a category of behavioral data, and performs feature extraction in a hierarchical learning process to separate the features into two outputs, including weight vectors of various behavioral categories and retained behavior data of main features.
7. The system for analyzing the correlation between the behavioral data and the medical outcome of claim 6, wherein the representation process of the feature representation module (302) is as follows: the DBM architecture model consists of a visible layer and t hidden layers and is used as the stack of t RBM architecture models;
the feature weight vector finally generated by the moduleFeatures of various behavior classes obtained by learning behavior data at K timesPerforming representation, namely representing the distribution of various types of behavior data at the Kth time node, and performing iterative training by taking K as a time step to obtain dynamic feature representation of various types of behavior data;
8. The system for analyzing the correlation between the behavioral data and the medical outcome of claim 3, wherein the correlation analysis module (303) uses an antagonistic neural network joint probability distribution estimation method for analysis, and comprises two models, wherein the first model takes the medical effect of the current time node as a label, the trained model predicts the medical effect of the next time node, the predicted result is compared with the corresponding actual medical effect, and when the model effect is reduced, the model is subjected to iterative optimization to obtain a reasonable dynamic prediction of the medical outcome; and the model II takes the final medical outcome as a label, the trained model predicts the final medical outcome of the new sample, and the analysis data is obtained by combining the feature weight vector output by the last module.
9. The system of claim 8, wherein the correlation module (303) comprises two models,
model-jointGAN 1 predicts the medical effect at the next time node T +1, with input vector YnextEach element of (a) indicates whether the medical outcome corresponding to the behavior data at the K moments is improved, if the improvement is stored as 1, otherwise, the improvement is 0, and the first prediction result of the model is yT+1If the actual result Y at the time of T +1 is observedT+1And yT+1If the model is not equal to the preset value, the model I is trained again;
model two jointGAN2 predicts the final medical outcome, with the input vector YfinalEach element of (a) corresponds to the final medical outcome of the sample, with the predicted result being yfinal。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010742790.6A CN111883247B (en) | 2020-07-29 | 2020-07-29 | Analysis system for correlation between behavior data and medical outcome |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010742790.6A CN111883247B (en) | 2020-07-29 | 2020-07-29 | Analysis system for correlation between behavior data and medical outcome |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111883247A true CN111883247A (en) | 2020-11-03 |
CN111883247B CN111883247B (en) | 2022-03-15 |
Family
ID=73202014
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010742790.6A Active CN111883247B (en) | 2020-07-29 | 2020-07-29 | Analysis system for correlation between behavior data and medical outcome |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111883247B (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101266710A (en) * | 2007-03-14 | 2008-09-17 | 中国科学院自动化研究所 | An all-weather intelligent video analysis monitoring method based on a rule |
CN104866727A (en) * | 2015-06-02 | 2015-08-26 | 陈宽 | Deep learning-based method for analyzing medical data and intelligent analyzer thereof |
CN109119137A (en) * | 2018-08-24 | 2019-01-01 | 腾讯科技(深圳)有限公司 | A kind of method for detecting abnormality, device, server and storage medium |
CN109740020A (en) * | 2018-12-26 | 2019-05-10 | 秒针信息技术有限公司 | Data processing method, device, storage medium and processor |
CN110277172A (en) * | 2019-06-27 | 2019-09-24 | 齐鲁工业大学 | A kind of clinical application behavior analysis system and its working method based on efficient negative sequence mining mode |
US20190333638A1 (en) * | 2015-11-16 | 2019-10-31 | Medecide Ltd. | Automated method and system for screening and prevention of unnecessary medical procedures |
CN110719441A (en) * | 2019-09-30 | 2020-01-21 | 傅程宏 | System and method for bank personnel behavior compliance early warning management |
-
2020
- 2020-07-29 CN CN202010742790.6A patent/CN111883247B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101266710A (en) * | 2007-03-14 | 2008-09-17 | 中国科学院自动化研究所 | An all-weather intelligent video analysis monitoring method based on a rule |
CN104866727A (en) * | 2015-06-02 | 2015-08-26 | 陈宽 | Deep learning-based method for analyzing medical data and intelligent analyzer thereof |
US20190333638A1 (en) * | 2015-11-16 | 2019-10-31 | Medecide Ltd. | Automated method and system for screening and prevention of unnecessary medical procedures |
CN109119137A (en) * | 2018-08-24 | 2019-01-01 | 腾讯科技(深圳)有限公司 | A kind of method for detecting abnormality, device, server and storage medium |
CN109740020A (en) * | 2018-12-26 | 2019-05-10 | 秒针信息技术有限公司 | Data processing method, device, storage medium and processor |
CN110277172A (en) * | 2019-06-27 | 2019-09-24 | 齐鲁工业大学 | A kind of clinical application behavior analysis system and its working method based on efficient negative sequence mining mode |
CN110719441A (en) * | 2019-09-30 | 2020-01-21 | 傅程宏 | System and method for bank personnel behavior compliance early warning management |
Non-Patent Citations (1)
Title |
---|
欧芳: "基于特征融合的行为识别", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
Also Published As
Publication number | Publication date |
---|---|
CN111883247B (en) | 2022-03-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108899064A (en) | Electronic health record generation method, device, computer equipment and storage medium | |
CN109117864A (en) | Coronary heart disease risk prediction technique, model and system based on heterogeneous characteristic fusion | |
CN109920501A (en) | Electronic health record classification method and system based on convolutional neural networks and Active Learning | |
CN111506723B (en) | Question-answer response method, device, equipment and storage medium | |
WO2020224433A1 (en) | Target object attribute prediction method based on machine learning and related device | |
CN111489800A (en) | Analysis method and system for identifying and storing medical record and report list images | |
Peng et al. | Sequential diagnosis prediction with transformer and ontological representation | |
CN113723513A (en) | Multi-label image classification method and device and related equipment | |
Viswan et al. | Explainable artificial intelligence in Alzheimer’s disease classification: A systematic review | |
CN113707323A (en) | Disease prediction method, device, equipment and medium based on machine learning | |
Zheng et al. | Autism classification based on logistic regression model | |
CN114724710A (en) | Emergency scheme recommendation method and device for emergency events and storage medium | |
Kampouraki et al. | e-Doctor: a web based support vector machine for automatic medical diagnosis | |
CN117391092B (en) | Electronic medical record multi-mode medical semantic alignment method based on contrast learning | |
CN114191665A (en) | Method and device for classifying man-machine asynchronous phenomena in mechanical ventilation process | |
CN114330482A (en) | Data processing method and device and computer readable storage medium | |
CN113855008A (en) | Wearable body fluid monitoring device and operation method thereof | |
CN111883247B (en) | Analysis system for correlation between behavior data and medical outcome | |
Schrills et al. | How to Answer Why--Evaluating the Explanations of AI Through Mental Model Analysis | |
CN116543917A (en) | Information mining method for heterogeneous time sequence data | |
CN114022698A (en) | Multi-tag behavior identification method and device based on binary tree structure | |
CN113688854A (en) | Data processing method and device and computing equipment | |
CN117524400B (en) | Patient data visualization management system and method | |
CN112365992A (en) | Medical examination data identification and analysis method based on NRS-LDA | |
Singh et al. | Malaria parasite recognition in thin blood smear images using squeeze and excitation networks |
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 |