CN116741411A - Intelligent health science popularization recommendation method and system based on medical big data analysis - Google Patents
Intelligent health science popularization recommendation method and system based on medical big data analysis Download PDFInfo
- Publication number
- CN116741411A CN116741411A CN202310725409.9A CN202310725409A CN116741411A CN 116741411 A CN116741411 A CN 116741411A CN 202310725409 A CN202310725409 A CN 202310725409A CN 116741411 A CN116741411 A CN 116741411A
- Authority
- CN
- China
- Prior art keywords
- science popularization
- data
- health science
- popularization
- intelligent
- 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.)
- Pending
Links
- 230000036541 health Effects 0.000 title claims abstract description 226
- 238000007405 data analysis Methods 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000003745 diagnosis Methods 0.000 claims abstract description 58
- 238000005516 engineering process Methods 0.000 claims abstract description 15
- 201000010099 disease Diseases 0.000 claims abstract description 9
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 9
- 238000011161 development Methods 0.000 claims abstract description 8
- 239000002775 capsule Substances 0.000 claims description 81
- 238000007689 inspection Methods 0.000 claims description 38
- 239000013598 vector Substances 0.000 claims description 33
- 230000006870 function Effects 0.000 claims description 29
- 230000002159 abnormal effect Effects 0.000 claims description 27
- 238000012216 screening Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 15
- 238000012360 testing method Methods 0.000 claims description 14
- 238000011144 upstream manufacturing Methods 0.000 claims description 10
- 238000004458 analytical method Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 7
- 238000010801 machine learning Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 239000008280 blood Substances 0.000 claims description 3
- 210000004369 blood Anatomy 0.000 claims description 3
- 210000002700 urine Anatomy 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 2
- 230000004927 fusion Effects 0.000 claims description 2
- 239000000463 material Substances 0.000 claims description 2
- 238000004891 communication Methods 0.000 abstract description 3
- 239000010410 layer Substances 0.000 description 77
- 238000010586 diagram Methods 0.000 description 8
- 230000002265 prevention Effects 0.000 description 6
- 230000004913 activation Effects 0.000 description 5
- 230000003068 static effect Effects 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 206010012601 diabetes mellitus Diseases 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 208000008589 Obesity Diseases 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000000712 assembly Effects 0.000 description 2
- 238000000429 assembly Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000007797 corrosion Effects 0.000 description 2
- 238000005260 corrosion Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 239000000428 dust Substances 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 239000011521 glass Substances 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000002203 pretreatment Methods 0.000 description 2
- 238000005507 spraying Methods 0.000 description 2
- 239000010959 steel Substances 0.000 description 2
- 239000002344 surface layer Substances 0.000 description 2
- 238000002604 ultrasonography Methods 0.000 description 2
- 240000004244 Cucurbita moschata Species 0.000 description 1
- 235000009854 Cucurbita moschata Nutrition 0.000 description 1
- 235000009852 Cucurbita pepo Nutrition 0.000 description 1
- 244000061176 Nicotiana tabacum Species 0.000 description 1
- 235000002637 Nicotiana tabacum Nutrition 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000002565 electrocardiography Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000003608 fece Anatomy 0.000 description 1
- 210000001503 joint Anatomy 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 230000013011 mating Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 235000020354 squash Nutrition 0.000 description 1
- 230000017105 transposition Effects 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
- G16H80/00—ICT specially adapted for facilitating communication between medical practitioners or patients, e.g. for collaborative diagnosis, therapy or health monitoring
-
- 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
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- 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
-
- 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
- G16H70/00—ICT specially adapted for the handling or processing of medical references
- G16H70/20—ICT specially adapted for the handling or processing of medical references relating to practices or guidelines
-
- 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
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Bioethics (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses an intelligent health science popularization recommendation method and system based on medical big data analysis, comprising the following steps: "Intelligent health science popularization platform", "intelligent health science popularization propaganda pavilion" and "intelligent health science popularization data recommendation model". The intelligent health science popularization platform is health science popularization propaganda software with authority science popularization number in an overall management area, based on disease diagnosis information of patients in various levels of hospitals, an off-line client terminal intelligent health science popularization propaganda pavilion is used as science popularization medium by utilizing a medical big data analysis technology, and scientific, customized and accurate pushing health science popularization and education data of each science popularization audience are automatically achieved through an intelligent health science popularization data recommendation model. Scientific series connection of science popularization data and science popularization audiences is achieved through medical big data analysis, a communication and communication platform between authoritative science popularization experts and the audiences is built, and the crossover type development of the healthy science popularization mode is guided.
Description
Technical Field
The invention belongs to the technical field of medical intelligent science popularization, and particularly relates to an intelligent health science popularization recommending method and system based on medical big data analysis.
Background
Health science popularization has become an important field of public attention, along with rapid development of information technology such as 'Internet+', 'artificial intelligence', 'big data', 'Internet of things', and the like and emergence of a plurality of new media platforms, more and more science popularization workers begin to work on innovative science popularization products by utilizing the new technology and the new platform, and the spanning development of science popularization work in the new era is promoted.
At present, the development of information technology is very different, popular science products and popular science data on the Internet are in explosive growth, and the traditional healthy popular science still adopts rough service modes such as 'news type', 'search type'. The duckling type science popularization method often enables science popularization audiences to be passively submerged in massive science popularization information, and can not quickly and accurately find out the healthy science popularization information truly suitable for the users according to the health conditions of the users. Meanwhile, the health science popularization data on the Internet are huge as the tobacco sea, and the source is five flowers and eight doors. Some denormal enterprises seek to search for high ranking on websites even through a payment competition mode, play against the "science popularization" and conduct illegal advertising. In addition, popular science audiences always passively accept popular science education, and have a difficult opportunity to raise personalized problems and communicate with authoritative specialists in close proximity. The popular science specialists are difficult to collect the most concerned health popular science problems of the audience, so that the popular science propaganda work can be developed in a targeted manner.
Through the analysis, the existing health science popularization mode has the following problems and defects:
1) And a contradiction between the healthy popular science content and popular science audience needs.
2) The health science popularization content lacks authority authentication and supervision.
3) There is a lack of a platform for effective communication between the popular science audience and the popular science specialist.
Disclosure of Invention
Aiming at the problems of the current health science popularization mode, the invention aims to solve the technical problem of providing an intelligent health science popularization recommending method and system based on medical big data analysis.
The technical scheme of the method of the invention is as follows:
in a first aspect, the present invention provides a medical big data analysis-based intelligent health science popularization recommendation method, the recommendation method comprising the following steps:
pre-training an ELMo model using the ICD-10 diagnostic dataset;
screening and sorting typical cases and health science popularization data, and constructing a typical case library and a health science popularization database; the typical case library includes a plurality of individual diagnostic data, test data, and typical cases of examination data;
pre-screening the inspection data and the check data, and eliminating the normal result to leave only an abnormal image and an abnormal inspection value;
the health science popularization database is a health science popularization data sample which comprises diagnosis tags and recommendation indexes corresponding to the diagnosis tags, and the same health science popularization data sample is provided with at least one diagnosis tag and recommendation index;
matching the diagnosis tag with the typical case from the aspect of health science popularization, so as to realize the many-to-many matching of the typical case and the health science popularization data sample, and obtain a matched typical case sample;
constructing an intelligent health science popularization data recommendation model, wherein the intelligent health science popularization data recommendation model is obtained by training a multimode-based improved capsule network by using matched typical case samples and is used for automatically classifying health science popularization data;
the improved capsule network based on the multiple modes comprises an upstream task and a downstream task, wherein the upstream task is composed of a pre-trained ELMo model, the pre-trained ELMo model is input into diagnostic data, and the pre-trained ELMo model is output into an embedded word vector;
the downstream task adopts a capsule network, the capsule network comprises a convolution layer ConV1, a main capsule layer, a convolution layer ConV2, a convolution capsule layer and a full-connection capsule layer which are sequentially connected, the input of the convolution layer ConV1 is an abnormal image, the image characteristics are obtained after the processing of the convolution layer ConV1, the embedded word vectors, the image characteristics and the abnormal inspection values are input into the main capsule layer, and the full-connection capsule layer is utilized to predict and output the recommendation probability of the health science popularization data.
According to the invention, research is carried out from two aspects of clinic and science popularization by constructing an authoritative science popularization expert team, a diagnosis tag and a recommendation index are marked for health science popularization data, and the diagnosis tag is matched with a typical case from the aspect of health science popularization.
The typical cases all come from real clinical data of a multi-family three-medical case system. The case data mainly relates to text data of case diagnosis, image data of examination items and numerical data of examination items.
Furthermore, the diagnostic data in the method of the invention all meet the coding standard of ICD-10. The image data of the examination item mainly relates to image data of electrocardiography, CT, MR, ultrasound, etc. The test items mainly relate to numerical data of test results such as blood, urine, feces, and the like.
The image data of the inspection item and the numerical data of the inspection item in the typical case are subjected to pre-screening treatment, the normal result is removed to leave only the abnormal image and the inspection numerical value, and then the abnormal image is subjected to pre-treatment such as normalization and denoising treatment.
Further, the multi-mode features can enter a convolution layer ConV2 after the main capsule layers are fused, and the number of the filters can be reduced to reduce feature dimension and training time and cost. The processed multi-modal features are input into a convolutional capsule layer, the capsule dimensions of the layer are matched with the classification of the health science popularization data and the corresponding typical medical records, and each capsule dimension represents the probability of a health science popularization data label.
The invention provides a medical big data analysis-based intelligent health science popularization recommendation system, which comprises an intelligent health science popularization platform, an intelligent health science popularization propaganda booth, an intelligent health science popularization data recommendation model, a typical case library, an abnormal data screening module, a health science popularization database and a matching module;
the intelligent health science popularization platform is used for comprehensively managing health science popularization propaganda software of authoritative science popularization numbers in an area, relies on a medical conjunct data center as a basis, takes a big data analysis technology as a grip, utilizes a big data analysis technology according to disease diagnosis information of patients in hospital at each level in the medical conjunct as a basis, takes an off-line client terminal as a science popularization medium, and automatically pushes health science popularization and education data for each science popularization audience scientifically, custom and accurately;
the intelligent health science popularization propaganda booth is hardware equipment for off-line health science popularization propaganda, and a touch display screen, an industrial control host, a medical insurance card reader, a laser printer, a sound box, a module indicator light, a sound box and a seat are integrated in the propaganda booth;
the abnormal data screening module is used for pre-screening the inspection data and the check data in the typical case library, and eliminating the normal result to only leave an abnormal image and an abnormal inspection value;
the matching module is used for marking a diagnosis tag and a recommendation index corresponding to the diagnosis tag on the health science popularization data sample in the health science popularization database, matching the diagnosis tag with the typical case from the health science popularization angle, realizing the many-to-many matching of the typical case and the health science popularization data sample, and obtaining a matched typical case sample;
the intelligent health science popularization data recommendation model is a customized health science popularization data recommendation model which is trained by performing supervised machine learning by adopting matched typical case samples.
The intelligent health science popularization platform comprises a science popularization module, a social module and an intelligent module; the method comprises three types of users, namely an administrator, a science popularization number and a science popularization user; wherein the administrator user has an account management function; the user with the science popularization number has the functions of authentication, checking, classifying and checking the health science popularization data and technical service; the "science popularization number" user is the medical and health institution, media and health science popularization expert of "big V certification", can release the authoritative health science popularization data through the science popularization number, the specific function includes uploading, modifying, deleting, setting up of the health science popularization data;
the popular science users are masses who want to accept popular science learning, and can customize and accurately push healthy popular science learning data meeting own needs for each audience through medical big data analysis;
the intelligent health science popularization platform supports development of science popularization social activities, and the social module comprises functions of forwarding, commenting, praying, subscribing and private letter of health science popularization information, and also comprises a 'science popularization circle' functional module, a 'hot search list' functional module, a 'super-speech' functional module, a 'science popularization forum' functional module and a 'science popularization knowledge competition' functional module;
the intelligent module comprises a chatgpt intelligent voice customer service function module, a VR science popularization function module, a 5G remote science popularization function module and a big data illness state analysis and prediction function module.
The labeled health science popularization data are ranked according to the correlation of the diagnosis labels, and are matched with typical cases in a 'many-to-many' mode, namely one typical case is matched with a plurality of health science popularization data, and one health science popularization data is matched with a plurality of typical cases.
The intelligent health science popularization data recommendation model is obtained by training a multimode-based improved capsule network by using matched typical case samples and is used for automatically classifying health science popularization data;
the improved capsule network based on the multiple modes comprises an upstream task and a downstream task, wherein the upstream task is composed of a pre-trained ELMo model, the pre-trained ELMo model is input into diagnostic data, and the pre-trained ELMo model is output into an embedded word vector;
the downstream task adopts a capsule network, the capsule network comprises a convolution layer ConV1, a main capsule layer, a convolution layer ConV2, a convolution capsule layer and a full-connection capsule layer which are sequentially connected, the input of the convolution layer ConV1 is an abnormal image, the image characteristics are obtained after the processing of the convolution layer ConV1, the embedded word vectors, the image characteristics and the abnormal inspection values are input into the main capsule layer, and the full-connection capsule layer is utilized to predict and output the recommendation probability of the health science popularization data.
The science popularization expert team performs auditing, screening, summarizing and diagnosing correlation to diagnose and recommend index marking according to the content of the healthy science popularization data, and completes pretreatment of the healthy science popularization data;
the preprocessed health science popularization data is accessed into a health science popularization database and is disclosed in an intelligent health science popularization platform;
when a science popularization user enters an intelligent health science popularization propaganda booth in an online manner through a medical insurance card swiping manner, the intelligent health science popularization platform calls the historical diagnosis and treatment information of the user in a medical conjunct data center through a health private network, and case data is fed back to the intelligent health science popularization platform;
the intelligent health science popularization platform calls an intelligent health science popularization data recommendation model to conduct multi-mode analysis on the case data of the user, and according to the multi-dimensional multi-label classification result of the intelligent health science popularization data recommendation model, proper health science popularization data TopN is screened from a health science popularization database, and finally customized and accurate pushing is conducted to science popularization users.
In a third aspect, the present invention provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the above method.
Compared with the prior art, the invention has the beneficial effects that:
the invention discloses an intelligent health science popularization recommendation system based on medical big data analysis, which constructs health science popularization products with an intelligent health science popularization platform (software), an intelligent health science popularization propaganda booth (equipment) and an intelligent health science popularization data recommendation model (method) as cores. Creatively proposes the ideas of customization and accurate health science popularization. An improved capsule network based on multiple modes is constructed, supervised deep learning is carried out on typical case samples, and an intelligent health science popularization data recommendation model is constructed. By taking the intelligent health science popularization propaganda platform as a support, authoritative media, institutions and experts are attracted to enter the platform, regional health science popularization resources are integrated, an intelligent health science popularization propaganda pavilion is established offline as a science popularization terminal, and a bridge between health science popularization knowledge and corresponding audiences is accurately built. Advanced informatization technologies such as medical big data analysis, machine learning, internet of things and the like are utilized, science popularization experience is improved, science popularization quality and efficiency are improved, and the crossing development of a healthy science popularization mode is promoted.
Compared with the conventional method for recommending the science popularization according to the diagnosis category, the recommendation model in the invention focuses on individuals of the science popularization audience. And (3) taking the diagnosis text, the inspection image and the inspection value in each medical record as training samples, carrying out fine granularity comprehensive analysis on the medical records from the angles of multiple modes and multiple dimensions, and matching the most suitable health science popularization data for each science popularization audience in a customized manner by combining with the health science popularization data marking in advance by the science popularization expert team. For example, individuals with the same medical history diagnosis may be more accurately and carefully recommended by model matching healthy science popularization data applicable to the same disease but different in severity due to differences in lesions on the inspection images and differences in the numerical values of inspection results.
Drawings
FIG. 1 is a flow chart of a method for recommending intelligent health science popularization based on medical big data analysis.
Fig. 2 is a schematic structural diagram of a modified capsule network based on multiple modes in the invention.
FIG. 3 is a schematic diagram of matching of healthy science popularization data with the science popularization audience in the present invention.
Fig. 4 is a functional architecture diagram of the intelligent health science popularization platform in the invention.
Fig. 5 is a schematic diagram of hardware architecture of the intelligent health science popularization propaganda booth according to the present invention.
Detailed Description
The following description of the embodiments of the present invention 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 invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.
It should be noted that, without conflict, the embodiments of the present invention and the methods, devices, and terminals in the embodiments may be combined with each other. The invention is further described below with reference to the drawings and specific examples, which are not intended to be limiting.
The invention provides a medical big data analysis-based intelligent health science popularization system, a method and equipment, wherein the intelligent health science popularization system mainly comprises three parts of an intelligent health science popularization platform (software), an intelligent health science popularization propaganda booth (equipment) and an intelligent health science popularization data recommendation model (method).
The intelligent health science popularization platform is health science popularization propaganda software of authority science popularization number in an overall management area, relies on a medical conjunct data center as a basis, uses a big data analysis technology as a gripper, uses a big data analysis technology according to disease diagnosis information of patients in hospital at all levels in the medical conjunct as a basis, uses an off-line client terminal intelligent health science popularization propaganda pavilion as a science popularization medium, and automatically pushes health science popularization propaganda and teaching data for each science popularization audience scientifically, custom and accurately.
Furthermore, the platform is a Web software application system based on a B/S architecture, and mainly comprises functional modules such as customized health science popularization propaganda, health science popularization social circles, intelligent AI interaction and the like. The platform can be divided into three types of users, namely an administrator, a science popularization number and a science popularization user according to the function roles. The administrator user has the functional rights of account management, science popularization number authentication, health science popularization data auditing, classification and checking, technical service and the like. The "science popularization number" user is generally the medical and health institution, media and health science popularization expert of "big V authentication". They can issue authoritative health science popularization data through science popularization numbers, and specific functions comprise uploading, modifying, deleting, top setting and the like of the health science popularization data. The popular science users are people who want to accept popular science learning, and the platform can customize and accurately push the healthy popular science learning data meeting the own needs for each audience through medical big data analysis. The platform supports development of science popularization social activities, comprises functions of forwarding, commenting, praying, subscribing, private letter and the like of healthy science popularization data, and has functional modules of science popularization circle, hot search list, superword, science popularization forum, science popularization knowledge competition and the like. In addition, the platform is provided with an intelligent interaction module, integrates a plurality of artificial intelligent models, and provides functional modules such as chatgpt intelligent voice customer service, VR family general, 5G remote family popularization big data disease analysis and prediction and the like.
The intelligent healthy science popularization propaganda booth is hardware equipment similar to the off-line healthy science popularization propaganda of a singing bar, the appearance of the intelligent healthy science popularization propaganda booth is a cuboid cabinet which is vertically placed, a whole frame is made of high-quality steel, and the intelligent healthy science popularization propaganda booth is firm and thick and cannot deform in a high-temperature and high-cold environment. The periphery of the cabinet adopts high-strength glass as a main material, the surface layer of the equipment adopts a high-end metal spraying technology, and the equipment is rust-proof, waterproof, durable and corrosion-resistant, can resist severe environmental interference, and has the characteristics of splash resistance, dust prevention, oil pollution prevention, static electricity prevention and the like. The propaganda pavilion is internally integrated with a touch display screen, an industrial control host, a medical insurance card reader, a laser printer, a sound box, a module indicator light, a sound box, a seat and other hardware equipment.
The flow of the intelligent health science popularization recommendation method based on medical big data analysis is shown in fig. 1, and the method mainly comprises the following steps:
step S1, screening and sorting typical cases and health science popularization data, and constructing a typical case library and a health science popularization database;
step S2, a "science popularization expert team" of the building authority develops discussion from two aspects of clinic and science popularization, marks a "diagnosis tag" and a "recommendation index" for the health science popularization data, and simultaneously matches the "diagnosis tag" with a typical case from the aspect of health science popularization;
step S3, randomly dividing the marked typical case sample into a training set, a testing set and a verification set;
s4, constructing an improved capsule network based on multiple modes, and performing supervised deep learning on the sample;
and S5, performing model parameter adjustment according to the classification performance on the verification set, evaluating the generalization capability of the model based on the matching effect on the test set, and finally training to obtain the intelligent health science popularization data recommendation model.
The detailed steps of this embodiment will be specifically described below in connection with the structural schematic diagram of the improved capsule network based on multiple modes in fig. 2.
The typical cases described in fig. 2 are all from real clinical data of the multi-family tri-hospital case system. The case data mainly relate to diagnosis data of case diagnosis, inspection data of inspection items and inspection data of inspection items, wherein the inspection data of the inspection items are image data, the inspection data of the inspection items are numerical data, and the diagnosis data are text data. The diagnostic data, the test data, and the examination data of a large number of individuals constitute a multi-modal case data set.
Furthermore, the diagnostic data in the method of the invention all meet the coding standard of ICD-10. The examination data of the examination item mainly relate to image data of electrocardio, CT, MR, ultrasound, etc. The test data of the test item mainly relates to numerical data of test results such as blood, urine, stool, and the like.
The health science popularization data are mainly obtained by collecting units such as medical universities, medical institutions, authoritative media, science popularization institutions and the like, and the content forms are mainly short videos of the health science popularization and can also comprise text forms.
The member of the science popularization expert team mainly comprises: medical college teachers, medical institution doctors, authoritative science popularization media editions, and science popularization society related expert group members.
The diagnosis label of the health science popularization data, namely inviting science popularization specialists to label the diagnosis label and the recommendation index for the health science popularization data by combining the specific content of the health science popularization data from the aspects of clinic and science popularization.
For example, a certain health science popularization data name is "how healthy a diabetic patient is losing weight? The health science popularization data is marked with diagnosis labels such as diabetes obesity-3, diabetes-2, obesity-1 and the like, wherein the tail number indicates the recommended index of the health science popularization data and the corresponding diseases.
Further, the annotated health science popularization data are ranked according to the recommendation index, and are matched with typical cases in a 'many-to-many' way. That is, one typical case matches multiple health science popularization data, and one health science popularization data matches multiple typical medical records. These "diagnostic tags", "recommended index" and "ranking results" will be important bases for supervised learning according to the present invention.
The improved capsule network based on the multiple modes comprises an upstream task and a downstream task, wherein the upstream task is composed of a pre-trained ELMo model, the ELMo model is input into diagnostic data, and the ELMo model is output into an embedded word vector; the downstream task adopts a capsule network, the embedded word vectors, the embedded image data and the embedded numerical data are respectively input into the capsule network, supervised machine learning is carried out, and an intelligent health science popularization data recommendation model is constructed. The model has the function of automatically classifying the health science popularization data according to the multi-mode case data.
The diagnostic data for the typical case diagnosis will be Word embedded (Word Embedding) using a pre-trained ELMo (Embedding from Language Models) model. The ELMo model is a pre-training language model for dynamically embedding words according to text context, and the internal double-layer Bi-LSTM (Bi-directional Long Short-Term Memory) pre-training model can dynamically adjust the input of words before and after a diagnosis sequence. The bi-layer bi-directional BiLSTM is actually an RNN model (Recurrent Neural Network).
Specifically, the ELMo in the invention adopts the ICD-10 diagnosis data set to conduct pre-training, so that the ELMo has the capability of identifying and diagnosing, and the standard diagnosis text in the ICD-10 diagnosis data set is beneficial to improving the efficiency of extracting the vector features of diagnosis information words in the typical medical record. Compared with static word2vec, ELMo is a bi-directional language model adopting a BiLSTM structure, and has better effect on word-ambiguous text data word embedding. The bi-directional language model comprises a forward model and a backward model, and a text T= [ T ] containing N words is given 1 ,t 2 ,…,t N ]The forward model needs to pass the previous word t 1 ,t 2 ,…,t k-1 ]Predicting the next word t k The formula is as follows:
wherein p is probability; t0 is the first word of the cut diagnostic text.
Then the model needs to predict the former word through the latter word, and the formula is as follows:
BiLSTM has a total L layer, for forward LSTM, each word t k-1 Is the last layer of output dynamic word vectorFor predicting the next word t k The method comprises the steps of carrying out a first treatment on the surface of the For backward LSTM, each word t k+1 Is the last layer of output dynamic word vectorFor predicting the previous word t k . The predictive procedure uses softmax, the objective function that BiLSTM needs to optimize is as follows:
wherein θ is x Representing word vector, θ when word is input s A softmax layer is shown for predicting the words before and after.Parameters representing forward LSTM for calculating +.> Parameters representing backward LSTM for calculating +.>
The bimtm layer number l=2 used in ELMo first trains a model on the ICD-10 diagnostic dataset and then in a subsequent task can output a word vector for each word based on the entered diagnostic data. The method for calculating the word vector by ELMo is as follows:
word vectors E (1), …, E (N) of words are first looked up from a static word vector table for input. ELMo uses the word vector generated by CNN-BIG-LSTM as input.
The word vectors E (1), … and E (N) are respectively input into a layer 1 forward LSTM and a layer 1 backward LSTM to obtain forward outputAnd backward output->
Output forward directionIs transmitted into a layer 2 forward LSTM to obtain a layer 2 forward outputOutput backward +.>Is transmitted into a layer 2 backward LSTM to obtain a layer 2 backward output
The word vector that the word i can ultimately derive includes E (i),with the BiLSTM of the L layer, 2L+1 word vectors can be finally obtained.
The emphasis point of word vectors of different layers in ELMo is usually different, CNN-BIG-LSTM word vectors adopted by an input layer can better encode part-of-speech information, LSTM of layer 1 can better encode syntax information, LSTM of layer 2 can better encode word semantic information.
The method for obtaining the word vector in ELMo is to carry out weighted fusion on the output of BiLSTM, and the specific formula is as follows:
where γ is a constant coefficient, allowing for word vector scaling of ELMo to increase the flexibility of the model.Is a weight coefficient normalized using softmax, i denotes the i-th word, j denotes the j-th layer.
According to the invention, text diagnosis information is input into a pre-trained ELMo model structure, 3 Word assemblies can be obtained in the ELMo model by each diagnosis vocabulary, each Word assembly is given a weight a, the weight can be obtained by learning, and three Word assemblies are integrated into one after summation according to the weights, so that the new feature of a capsule network is improved as a subsequent input.
The image data of the inspection item and the numerical data of the inspection item in the typical case are subjected to pre-screening treatment, the normal result is removed to leave only the abnormal image and the abnormal inspection numerical value, and then the abnormal image is subjected to pre-treatment such as normalization and denoising treatment.
The capsule network is a convolutional neural network which is built by adopting a pyrach programming language and has a plurality of capsule layer structures, and the network structure is a standard convolutional layer (ConV 1), a main capsule layer, a standard convolutional layer (ConV 2), a convolutional capsule layer and a fully-connected capsule layer which are sequentially connected in 5 layers as shown in figure 2.
The input of the downstream task can be divided into 2 branches, one branch is that word vectors and abnormal test data generated by ELMo are directly input into a main capsule layer, the other branch is that the preprocessed test data is firstly input into a standard convolution layer (a convolution layer ConV 1) to perform feature extraction to obtain image features, then the image features are in butt joint with the main capsule layer, the final image features, the word vectors and the abnormal test data can be fused with multi-mode data in the main capsule layer, and the output of the main capsule layer is classified by a standard convolution layer (a convolution layer ConV 2) and a convolution capsule layer through a softmax function. The classification results are a plurality of diagnosis labels and recommendation indexes, the models are ranked again according to the recommendation indexes, and finally the healthy science popularization data corresponding to the recommendation indexes TopN are recommended to the science popularization audiences, so that the personalized matching of the healthy science popularization data and the multi-dimensional multi-labels of the science popularization audiences is realized.
The main capsule layer is different from the traditional convolutional layer, and uses vector capsules to replace neurons in a convolutional neural network, so that scalar features are converted into vector features, and word sequence and semantics of text features are reserved to the greatest extent. Specifically, a step is added when the weighted summation is carried out, and the specific working principle can be divided into the following 4 steps:
1) Low-level features u of the main capsule layer i Multiplying by weight matrix w between high-level features ij Obtaining new high-level characteristics u j|i The formula is:
u i *w ij =u j|i (5)
2) Will be high-level features u j|i Multiplying by weight c j|i Wherein c j|i Is determined by dynamic routing.
3) For all u j|i *c j|i Summing to obtain a vector S j 。
4) Finally using the compressed activation function squarash to make S j Conversion to v j 。
The capsule network uses a compressed activation function square to replace a RelU function used by a traditional convolutional neural network activation layer, and the formula is as follows:
the activation function both ensures that the data is between 0-1 and also preserves the direction of the vector.
In addition, the main capsule layer of the capsule network adopts a dynamic routing mode to replace the traditional pooling operation for feature selection. I.e. the input features are clustered, wherein the more similar features, the stronger the features.
The specific process of dynamic routing of the main capsule layer can be expressed as follows:
1) The initialization weight is 0:
2) T cycles were performed: for r in range (1, 2, …, T).
3) The coupling coefficient is obtained in the cycle:
4) The activation vector is obtained in the loop: v r =squash(c 1r u 1 +c 2r u 2 +c 3r u 3 )。
5) The weights are updated sequentially in the loop: b 1(r+1) =b 1r +v r u 1 、b 2(r+1) =b 2r +v r u 2 、b 3(r+1) =b 3r +v r u 3 。
Furthermore, the multi-mode features can enter the convolution layer after the main capsule layer is fused, so that the number of filters can be reduced to reduce feature dimension, training time and training cost.
Further, the multi-modal feature processed by the main capsule layer and the standard convolution layer (convolution layer ConV 2) is input into the convolution capsule layer, the capsule dimension of the layer is the same as the number of diagnostic tags of the health science popularization data, and each capsule dimension represents the probability of one diagnostic tag of the health science popularization data. The capsule dimension of the convolution capsule layer is related to the number of typical cases and the respective diagnostic classifications; a health science popularization data corresponds to a plurality of diagnosis labels and a plurality of recommendation indexes. The dimensions herein will vary with the dimension of the "diagnostic tag" corresponding to the health science popularization data, e.g., a health science popularization data having 4 "diagnostic tags" would correspond to a capsule dimension of 4.
Further, the convolved capsule layers may be flattened into a list of capsules connected to the fully connected capsule layers. The full-connection capsule layer comprises a full-connection layer and a softmax classifier, and the output formula of the full-connection layer is as follows:
h W,b (x)=f(W T x+b) (7)
where x is the input of the neuron, h W,b (x) For output, W T And for transposition of W, the output nodes of the full connection layer are connected into a softmax classifier to carry out final probability prediction so as to finish classification of typical medical records.
And sorting the classification results according to the recommendation index, setting the ranking number of the TopN, and recommending the related health science popularization data with the top ranking.
Meanwhile, the diagnosis label of the classification result corresponds to the corresponding health science popularization data.
Further, model parameter adjustment is carried out according to the performance of the verification set, the generalization capability of the model is estimated based on the classification effect of the test set, and the intelligent health science popularization data recommendation model with good performance is obtained by adopting a cross verification method.
FIG. 3 is a schematic diagram showing the matching of the healthy science popularization data with the science popularization audience in the present invention.
Firstly, an authoritative science popularization expert team carries out auditing, screening, summarizing and diagnosing relevance according to the content of the healthy science popularization data to carry out diagnosis tag and recommendation index marking, and the pretreatment work of the healthy science popularization data is completed.
The preprocessed health science popularization data (marked with diagnosis tags and recommendation indexes) is then accessed into a health science popularization database and disclosed on the platform.
When a science popularization user enters the intelligent health science popularization propaganda pavilion in an online manner through a medical insurance card swiping manner, the intelligent health science popularization platform can call the historical diagnosis and treatment information of the user in the medical conjuncted data center through the health private network, and case data is fed back to the intelligent health science popularization platform.
Further, the intelligent health science popularization platform calls an intelligent health science popularization data recommendation model to conduct multi-mode analysis on the case data of the user, and according to the multi-dimensional multi-label classification result of the intelligent health science popularization data recommendation model, proper health science popularization data TopN is screened from a health science popularization database, and finally customized and accurate pushing is conducted on the corresponding science popularization users.
FIG. 4 is a functional architecture diagram of the intelligent health science popularization platform according to the present invention.
In this embodiment, the "intelligent health science popularization platform" mainly includes: science popularization module, social module (healthy science popularization social circle) and intelligent module.
The science popularization module covers the following functions:
1) Uploading, auditing, classifying and pushing the health science popularization data.
2) Science popularization propaganda and science popularization originality.
3) Popular science contests and popularity ranking.
The social module encompasses the following functions:
1) Popular science number "big V" authentication.
2) The healthy science popularization data is transferred, evaluated and praise.
3) Private letter, subscription between science popularization users and science popularization numbers, and a fan system.
4) Science popularization forum, expert overtaking, etc.
The intelligent module covers the following functions:
1) Intelligent human-machine interactions, such as chatgpt module embedding.
2) VR science popularization experience.
3) 5G remote popular science questions.
4) Big data disease analysis and prediction.
Fig. 5 is a schematic diagram of a hardware architecture of the intelligent health science popularization propaganda booth according to the present invention.
In this embodiment, the hardware devices of the "intelligent health science popularization propaganda booth" as the science popularization terminal may be divided into three parts including a cabinet housing, a client terminal and related mating devices.
Specifically, the appearance of propaganda pavilion is the cuboid rack of standing vertically, and whole machine frame adopts high-quality steel preparation, and is firm thick, can not warp under the environment of high temperature high and cold. The high-strength glass baffles are adopted around the cabinet, the high-end metal plastic spraying technology is adopted on the surface layer of the equipment, so that the equipment is rust-proof, waterproof, durable and corrosion-resistant, can resist severe environmental interference, and has the characteristics of splash resistance, dust prevention, oil pollution prevention, static electricity prevention and the like.
The client terminal in the propaganda booth integrates a touch display screen, an industrial control host, a laser printer and a sound box, and is equipped with medical insurance card swiping device, a seat, an indicator light, a power module and other matched equipment.
The intelligent health science popularization propaganda booth is used as a client terminal for falling to the ground of a science popularization product line, can support the realization of each functional module of an intelligent health science popularization platform, is used as a medium for the popularization and teaching of health science popularization, and provides comfortable, quiet and airtight science popularization study space for science popularization audiences.
The intelligent health science popularization system, method, equipment and terminal based on big data analysis creatively propose the concept of customizing and accurately health science popularization and promote the reform of the traditional 'extensive' science popularization mode. Depending on the medical conjunct data center, the bridge between the healthy science popularization knowledge and the corresponding audience is accurately built by utilizing the medical big data analysis technology. The regional health science popularization resource is integrated, an authoritative intelligent health science popularization propaganda platform is constructed, authoritative media, institutions and specialists are attracted to the resident platform, and the online and offline health science popularization activities are comprehensively managed. An intelligent health science popularization propaganda pavilion is established offline as a science popularization terminal, and science popularization experience is improved and science popularization quality and efficiency are improved through advanced informatization technologies such as artificial intelligence, the Internet of things, VR and 5G.
The preferred embodiments of the present invention have been described in detail, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the spirit of the present invention, and the various changes are included in the scope of the present invention.
The invention is applicable to the prior art where it is not described.
Claims (10)
1. The intelligent health science popularization recommending method based on medical big data analysis is characterized by comprising the following steps of:
pre-training an ELMo model using the ICD-10 diagnostic dataset;
screening and sorting typical cases and health science popularization data, and constructing a typical case library and a health science popularization database; the typical case library includes a plurality of individual diagnostic data, test data, and typical cases of examination data;
pre-screening the inspection data and the check data, and eliminating the normal result to leave only an abnormal image and an abnormal inspection value;
the health science popularization database is a health science popularization data sample which comprises diagnosis tags and recommendation indexes corresponding to the diagnosis tags, and the same health science popularization data sample is provided with at least one diagnosis tag and recommendation index;
matching the diagnosis tag with the typical case from the aspect of health science popularization, so as to realize the many-to-many matching of the typical case and the health science popularization data sample, and obtain a matched typical case sample;
constructing an intelligent health science popularization data recommendation model, wherein the intelligent health science popularization data recommendation model is obtained by training a multimode-based improved capsule network by using matched typical case samples and is used for automatically classifying health science popularization data;
the improved capsule network based on the multiple modes comprises an upstream task and a downstream task, wherein the upstream task is composed of a pre-trained ELMo model, the pre-trained ELMo model is input into diagnostic data, and the pre-trained ELMo model is output into an embedded word vector;
the downstream task adopts a capsule network, the capsule network comprises a convolution layer ConV1, a main capsule layer, a convolution layer ConV2, a convolution capsule layer and a full-connection capsule layer which are sequentially connected, the input of the convolution layer ConV1 is an abnormal image, the image characteristics are obtained after the processing of the convolution layer ConV1, the embedded word vectors, the image characteristics and the abnormal inspection values are input into the main capsule layer, and the full-connection capsule layer is utilized to predict and output the recommendation probability of the health science popularization data.
2. The intelligent health science popularization recommendation method based on medical big data analysis according to claim 1, wherein the main capsule layer adopts a dynamic routing mode to perform feature selection so as to realize multi-mode feature fusion; the capsule dimension of the convolution capsule layer is the same as the number of the diagnosis tags of the health science popularization data, namely, each capsule dimension represents the probability of one diagnosis tag of the health science popularization data.
3. The intelligent health science popularization recommendation method based on medical big data analysis according to claim 1, wherein the diagnosis tag and the recommendation index are marked by a science popularization expert team, the science popularization expert team performs discussion in terms of clinical and science popularization, and the diagnosis tag and the recommendation index are marked for health science popularization data; the typical cases are all real clinical data from a plurality of trimethyl hospital case systems; the real clinical data mainly relate to diagnosis data of case diagnosis, inspection data of inspection items and inspection data of inspection items, wherein the inspection data of the inspection items are image data, the inspection data of the inspection items are numerical data, and the diagnosis data are text data; the examination item mainly relates to numerical data of blood, urine and stool examination results, and the image data of the examination item mainly relates to electrocardiographic, CT, MR and ultrasonic image data.
4. The medical big data analysis-based intelligent health science popularization recommendation method according to claim 3, wherein the science popularization expert team members mainly comprise: medical college teachers, medical institution doctors, authoritative science popularization media editions and science popularization society related expert group members;
from the aspects of clinic and science popularization, the members of the science popularization expert team label the disease diagnosis which the corresponding science popularization audience generally should have for the healthy science popularization data by combining the specific content of the healthy science popularization data, and comprehensively calibrate the recommended index corresponding to the healthy science popularization data.
5. The intelligent health science popularization recommending system based on the medical big data analysis is characterized by comprising an intelligent health science popularization platform, an intelligent health science popularization propaganda pavilion, an intelligent health science popularization data recommending model, a typical case library, an abnormal data screening module, a health science popularization database and a matching module;
the intelligent health science popularization platform is used for comprehensively managing health science popularization propaganda software of authoritative science popularization numbers in an area, relies on a medical conjunct data center as a basis, takes a big data analysis technology as a grip, utilizes a big data analysis technology according to disease diagnosis information of patients in hospital at each level in the medical conjunct as a basis, takes an off-line client terminal as a science popularization medium, and automatically pushes health science popularization and education data for each science popularization audience scientifically, custom and accurately;
the intelligent health science popularization propaganda booth is hardware equipment for off-line health science popularization propaganda, and a touch display screen, an industrial control host, a medical insurance card reader, a laser printer, a sound box, a module indicator light, a sound box and a seat are integrated in the propaganda booth;
the abnormal data screening module is used for pre-screening the inspection data and the check data in the typical case library, and eliminating the normal result to only leave an abnormal image and an abnormal inspection value;
the matching module is used for marking a diagnosis tag and a recommendation index corresponding to the diagnosis tag on the health science popularization data sample in the health science popularization database, matching the diagnosis tag with the typical case from the health science popularization angle, realizing the many-to-many matching of the typical case and the health science popularization data sample, and obtaining a matched typical case sample;
the intelligent health science popularization data recommendation model is a customized health science popularization data recommendation model which is trained by performing supervised machine learning by adopting matched typical case samples.
6. The intelligent health science popularization recommendation system based on the medical big data analysis of claim 5,
the intelligent health science popularization platform comprises a science popularization module, a social module and an intelligent module; the method comprises three types of users, namely an administrator, a science popularization number and a science popularization user; wherein the administrator user has an account management function; the user with the science popularization number has the functions of authentication, checking, classifying and checking the health science popularization data and technical service; the "science popularization number" user is the medical and health institution, media and health science popularization expert of "big V certification", can release the authoritative health science popularization data through the science popularization number, the specific function includes uploading, modifying, deleting, setting up of the health science popularization data;
the popular science users are masses who want to accept popular science learning, and can customize and accurately push healthy popular science learning data meeting own needs for each audience through medical big data analysis;
the intelligent health science popularization platform supports development of science popularization social activities, and the social module comprises functions of forwarding, commenting, praying, subscribing and private letter of health science popularization information, and also comprises a 'science popularization circle' functional module, a 'hot search list' functional module, a 'super-speech' functional module, a 'science popularization forum' functional module and a 'science popularization knowledge competition' functional module;
the intelligent module comprises a chatgpt intelligent voice customer service function module, a VR science popularization function module, a 5G remote science popularization function module and a big data illness state analysis and prediction function module.
7. The medical big data analysis based intelligent health science popularization recommendation system according to claim 5, wherein the labeled health science popularization data are ranked according to the correlation of the diagnosis labels and are matched with the typical cases in a 'many-to-many' way, namely, one typical case is matched with a plurality of health science popularization data, and one health science popularization data is matched with a plurality of typical cases.
8. The medical big data analysis-based intelligent health science popularization recommendation system according to claim 5, wherein the intelligent health science popularization recommendation model is obtained by training a multi-mode-based improved capsule network by using matched typical case samples and is used for automatically classifying health science popularization materials;
the improved capsule network based on the multiple modes comprises an upstream task and a downstream task, wherein the upstream task is composed of a pre-trained ELMo model, the pre-trained ELMo model is input into diagnostic data, and the output is an embedded word vector, namely a diagnostic classification result;
the downstream task adopts a capsule network, the capsule network comprises a convolution layer ConV1, a main capsule layer, a convolution layer ConV2, a convolution capsule layer and a full-connection capsule layer which are sequentially connected, the input of the convolution layer ConV1 is an abnormal image, the image characteristics are obtained after the processing of the convolution layer ConV1, the embedded word vectors, the image characteristics and the abnormal inspection values are input into the main capsule layer, and the full-connection capsule layer is utilized to predict and output the recommendation probability of the health science popularization data.
9. The medical big data analysis based intelligent health science popularization recommendation system of claim 6,
the science popularization expert team performs auditing, screening, summarizing and diagnosing correlation to diagnose and recommend index marking according to the content of the healthy science popularization data, and completes pretreatment of the healthy science popularization data;
the preprocessed health science popularization data is accessed into a health science popularization database and is disclosed in an intelligent health science popularization platform;
when a science popularization user enters an intelligent health science popularization propaganda booth in an online manner through a medical insurance card swiping manner, the intelligent health science popularization platform calls the historical diagnosis and treatment information of the user in a medical conjunct data center through a health private network, and case data is fed back to the intelligent health science popularization platform;
the intelligent health science popularization platform calls an intelligent health science popularization data recommendation model to conduct multi-mode analysis on the case data of the user, and according to the multi-dimensional multi-label classification result of the intelligent health science popularization data recommendation model, proper health science popularization data TopN is screened from a health science popularization database, and finally customized and accurate pushing is conducted to science popularization users.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the steps of the method according to claims 1-4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310725409.9A CN116741411A (en) | 2023-06-19 | 2023-06-19 | Intelligent health science popularization recommendation method and system based on medical big data analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310725409.9A CN116741411A (en) | 2023-06-19 | 2023-06-19 | Intelligent health science popularization recommendation method and system based on medical big data analysis |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116741411A true CN116741411A (en) | 2023-09-12 |
Family
ID=87909416
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310725409.9A Pending CN116741411A (en) | 2023-06-19 | 2023-06-19 | Intelligent health science popularization recommendation method and system based on medical big data analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116741411A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116932920A (en) * | 2023-09-18 | 2023-10-24 | 青岛理工大学 | Accurate healthy science popularization data recommendation method based on big data |
CN117112834A (en) * | 2023-10-24 | 2023-11-24 | 苏州元脑智能科技有限公司 | Video recommendation method and device, storage medium and electronic device |
CN117708437A (en) * | 2024-02-05 | 2024-03-15 | 四川日报网络传媒发展有限公司 | Recommendation method and device for personalized content, electronic equipment and storage medium |
CN117954068A (en) * | 2024-03-27 | 2024-04-30 | 上海交通大学医学院附属仁济医院 | Science popularization information initiative matching and pushing system based on Internet hospital |
-
2023
- 2023-06-19 CN CN202310725409.9A patent/CN116741411A/en active Pending
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116932920A (en) * | 2023-09-18 | 2023-10-24 | 青岛理工大学 | Accurate healthy science popularization data recommendation method based on big data |
CN116932920B (en) * | 2023-09-18 | 2023-12-12 | 青岛理工大学 | Accurate healthy science popularization data recommendation method based on big data |
CN117112834A (en) * | 2023-10-24 | 2023-11-24 | 苏州元脑智能科技有限公司 | Video recommendation method and device, storage medium and electronic device |
CN117112834B (en) * | 2023-10-24 | 2024-02-02 | 苏州元脑智能科技有限公司 | Video recommendation method and device, storage medium and electronic device |
CN117708437A (en) * | 2024-02-05 | 2024-03-15 | 四川日报网络传媒发展有限公司 | Recommendation method and device for personalized content, electronic equipment and storage medium |
CN117708437B (en) * | 2024-02-05 | 2024-04-16 | 四川日报网络传媒发展有限公司 | Recommendation method and device for personalized content, electronic equipment and storage medium |
CN117954068A (en) * | 2024-03-27 | 2024-04-30 | 上海交通大学医学院附属仁济医院 | Science popularization information initiative matching and pushing system based on Internet hospital |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Luo et al. | A critical review of state‐of‐the‐art chatbot designs and applications | |
US11748555B2 (en) | Systems and methods for machine content generation | |
Xing et al. | Dropout prediction in MOOCs: Using deep learning for personalized intervention | |
US20230252224A1 (en) | Systems and methods for machine content generation | |
CN116741411A (en) | Intelligent health science popularization recommendation method and system based on medical big data analysis | |
Kaklauskas et al. | Intelligent decision support systems | |
Yan et al. | Online doctor recommendation with convolutional neural network and sparse inputs | |
Garbuio et al. | An algorithm for designing value propositions in the IoT space: Addressing the challenges of selecting the initial class in reference class forecasting | |
Lin et al. | Research on sound imagery of electric shavers based on kansei engineering and multiple artificial neural networks | |
Velásquez et al. | Expanding the possibilities of deliberation: The use of data mining for strengthening democracy with an application to education reform | |
Muangnak et al. | The neural network conversation model enables the commonly asked student query agents | |
Liu et al. | Brain-inspired multimodal learning based on neural networks | |
Wang et al. | Construction of a novel production develop decision model based on text mined | |
Naseem et al. | Graph learning with label attention and hyperbolic embedding for temporal event prediction in healthcare | |
Zhao et al. | Research on emotion-embedded design flow based on deep learning technology | |
Roy et al. | MultiMICS: a contextual multifaceted intelligent multimedia information fusion paradigm | |
Kurylets et al. | Threat modeling in RPA-Based systems | |
Akila | Disease Identification using Machine Learning and NLP | |
Marzullo | A service view of “smart” ACAP: the IBM Watson case | |
Kaur et al. | Management of Data in AI Age | |
Rajkumar et al. | Intelligent Chatbot for Hospital Recommendation System | |
Alanya-Beltran | Personalized Learning Recommendation System in E-learning Platforms Using Collaborative Filtering and Machine Learning | |
Kalpana et al. | Hybrid optimisation enabled deep learning for sentiment rating prediction towards text summarisation and question answering system | |
Huang et al. | Employing feature engineering strategies to improve the performance of machine learning algorithms on echocardiogram dataset | |
Hechler et al. | Relevant ML and DL Concepts |
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 |