CN111128375B - Tibetan medicine diagnosis auxiliary device based on multi-label learning - Google Patents

Tibetan medicine diagnosis auxiliary device based on multi-label learning Download PDF

Info

Publication number
CN111128375B
CN111128375B CN202010026148.8A CN202010026148A CN111128375B CN 111128375 B CN111128375 B CN 111128375B CN 202010026148 A CN202010026148 A CN 202010026148A CN 111128375 B CN111128375 B CN 111128375B
Authority
CN
China
Prior art keywords
case
classifier
label
classifiers
training
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.)
Active
Application number
CN202010026148.8A
Other languages
Chinese (zh)
Other versions
CN111128375A (en
Inventor
李巧勤
巩小强
刘勇国
杨尚明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN202010026148.8A priority Critical patent/CN111128375B/en
Publication of CN111128375A publication Critical patent/CN111128375A/en
Application granted granted Critical
Publication of CN111128375B publication Critical patent/CN111128375B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biomedical Technology (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Pathology (AREA)
  • Databases & Information Systems (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Medical Treatment And Welfare Office Work (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

The invention belongs to the field of national medicine aided decision making, and particularly provides a Tibetan medicine diagnosis assisting device based on multi-label learning, which is used for solving some problems in the traditional Tibetan medicine legislation process: namely inaccuracy, difference and lack of objective explanation of diagnosis results, the invention can provide auxiliary decision support for the diagnosis and treatment process of Tibetan doctors, and has the following advantages: 1) constructing a characteristic view: dynamically updating the feature view through continuous iteration of the training process; 2) and (3) label transmission: the most reliable samples are transmitted between the 3 classifiers in pairs, the classifiers are optimized, and the classification accuracy is improved; 3) selecting an optimal classification result: and selecting the optimal prediction result by using a method combining vector splicing and weight distribution.

Description

Tibetan medicine diagnosis auxiliary device based on multi-label learning
Technical Field
The invention belongs to the field of national medicine auxiliary decision making, and particularly relates to a Tibetan medicine diagnosis auxiliary device based on multi-label learning.
Background
Tibetan medicine is an important component of the medical treasury of China, has a long history of more than three thousand years and makes a great contribution to the prosperity of the Tibetan nationality and the whole Chinese nation; the scientific research and the inheritance of the Tibetan medicine are paid attention. In the clinical trial of Tibetan medicine, the physician mainly carries out the diagnosis and treatment of the patient according to the following 3 steps: 1) syndrome differentiation: doctors determine the syndrome type of patients according to the etiology and pathogenesis, 2) legislation: determining a treatment method according to the syndrome of the patient, and 3) composing a formula: making a prescription according to syndrome type and treatment; the treatment method is the basis of the clinical prescription sending, the method is established according to the syndrome, the method is established according to the method, and the result of the establishment directly influences the clinical curative effect. In recent years, many researchers use association rules, cluster analysis, topic models and other methods to mine effective Tibetan medical treatment rules from a large number of clinical cases, but all the methods mainly focus on syndrome differentiation and formula rule analysis. The research on the Tibetan medicine legislative rule is still in the traditional manual method stage, namely, doctors predict the therapeutic method according to long-term accumulated self experience.
When the traditional manual method is used for predicting the Tibetan medicine method, the accuracy of diagnosis results is reduced mainly by depending on personal experiences of doctors, different doctors may obtain different methods, and the diagnosis results obtained according to the personal experiences of the doctors are lack of objective explanation.
Disclosure of Invention
The invention aims to provide a Tibetan medicine diagnosis auxiliary device based on multi-label learning aiming at some problems in the traditional Tibetan medicine legislation process, namely inaccuracy, difference and lack of objectivity explanation of diagnosis results, and provides auxiliary decision support for the diagnosis and treatment process of Tibetan doctors.
In order to achieve the purpose, the invention adopts the technical scheme that:
a Tibetan medicine diagnosis assisting device based on multi-label learning, which comprises an input device for inputting symptoms, syndrome outlines and syndromes, and a classifier for receiving the input symptoms, syndrome outlines and syndromes, wherein the classifier is obtained by training the following processes:
step 1: constructing a data set comprising: m chronic kidney disease cases with treatment labels are called as a marked case set for short, and N chronic kidney disease cases without treatment labels are called as a unmarked case set for short;
step 2: training initial classifier based on feature view
Step 2.1: initializing a characteristic view: randomly extracting 1 case from the marked case set, and taking corresponding symptoms, syndrome and machine summaries and syndromes as 3 initialized feature views respectively;
step 2.2: training an initial classifier: training a classifier based on symptom features, a classifier based on syndrome machine summary features and a classifier based on syndrome features on 3 feature views by using an ML-KNN method based on a labeled case set, and sequentially labeling the classifiers as classifiers 1,2 and 3; and updating the corresponding 3 feature views after the training process is completed each time; repeating the training until all the marked cases are trained;
and step 3: optimizing classifier
Step 3.1: and (4) predicting a classification result: using 3 classifiers obtained by training to carry out classification prediction on the unmarked case set to obtain a prediction result
Figure BDA0002362538650000021
Figure BDA0002362538650000022
Wherein v represents a classifier number,
Figure BDA0002362538650000023
a probability vector representing the classifier v predicting case j,
Figure BDA0002362538650000024
represents the probability that the label of case j is t, where t is 1,2,.., K is the total number of therapeutic (labels) given a classifier v;
step 3.2: updating the prediction result:
Figure BDA0002362538650000025
Figure BDA0002362538650000026
wherein,
Figure BDA0002362538650000027
mark representing case j under the condition that classifier is v after updatingThe probability of the signature being t,
Figure BDA0002362538650000028
a probability vector representing the updated classifier v prediction case j;
Figure BDA0002362538650000029
represent the probability that case j is not associated with label t:
Figure BDA00023625386500000210
Figure BDA00023625386500000211
representing the probability that case j is associated with label t:
Figure BDA00023625386500000212
C(ta,tb) Indicates the label taAnd a label tbThe correlation of (a):
Figure BDA00023625386500000213
C(ta,tb)≠C(tb,ta)
wherein,
Figure BDA0002362538650000031
representing simultaneous inclusion of labels t in a marked case setaAnd tbThe number of cases (a) of (b),
Figure BDA0002362538650000032
representing only labels t in a marked case setbThe number of cases of (c);
step 3.3: calculating the global reliability of the prediction case j:
Figure BDA0002362538650000033
wherein Hv(j, t) represents the reliability of case j on label t:
Figure BDA0002362538650000034
step 3.4: performing descending order according to the global reliability, and selecting the first n cases as a reliable case set BvTransferring reliable case sets among the 3 classifiers, removing the reliable case sets corresponding to the unmarked case sets, adding the reliable case sets into the marked case sets to form new marked case sets, and training each classifier again on the new marked case sets by using an ML-KNN method;
step 3.5: repeating the steps 3.1-3.4 until the unmarked case set is empty; obtaining 3 final trained classifiers;
and 4, step 4: outputting an optimal classification result:
Figure BDA0002362538650000035
where α, β, γ are preset weights for classifiers 1,2, 3, respectively, and α + β + γ is 1.
The invention has the beneficial effects that:
the invention provides a Tibetan medicine diagnosis auxiliary device based on multi-label learning, which provides auxiliary decision support for the diagnosis and treatment process of Tibetan doctors; has the following advantages:
1) constructing a characteristic view: dynamically updating the feature view through continuous iteration of the training process;
2) and (3) label transmission: the most reliable samples are transmitted between the 3 classifiers in pairs, the classifiers are optimized, and the classification accuracy is improved;
3) selecting an optimal classification result: and selecting the optimal prediction result by using a method combining vector splicing and weight distribution.
Drawings
Fig. 1 is a schematic diagram of a training process of a classifier according to an embodiment of the present invention.
FIG. 2 is a diagram of a model structure of a classifier according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
For the purpose of facilitating an understanding of the present invention, reference will first be made to the basic definitions to which the invention relates:
symptoms are: the disease symptoms and clinical manifestations of the disease, such as sweating, dizziness, tinnitus, fever and the like, are the original basis for judging the disease;
the syndrome: the disease diagnosis and treatment method comprehensively analyzes various symptoms and summarizes the pathological aspects of the etiology, the disease location, the disease nature and the like at a certain stage in the process of the occurrence and the development of the diseases;
the summary of the certificate machine: the method refers to the summary of etiology and pathogenesis, wherein the etiology refers to the cause of a disease, and the pathogenesis refers to the mechanism of occurrence, development, change and fate of the disease;
therapeutic method: it refers to the basic method of treating diseases, such as warming the stomach to dispel cold.
In the embodiment, the medical record of chronic kidney disease is taken as data to carry out the recommendation and research of Tibetan medicine and treatment method, and the medical record data mainly comprises four parts of symptoms, syndrome and machine summary, syndrome and treatment method; in the chronic kidney disease medical record data set, most medical records contain at least 2 treatments. Therefore, the Tibetan medicine therapeutic method prediction is treated as a multi-label classification problem, namely, the Tibetan medicine therapeutic method prediction is classified and predicted by taking a therapeutic method as a label and taking symptoms, syndrome outlines and syndromes as 3 different attribute characteristics.
The embodiment provides a Tibetan medicine diagnosis auxiliary device based on multi-label learning, which comprises an input device for inputting symptoms, a syndrome summary and syndromes, and a classifier for receiving the input symptoms, the syndrome summary and the syndromes, wherein the training process of the classifier is as shown in fig. 1, and specifically comprises the following steps:
step 1: constructing a data set comprising: m chronic kidney disease cases with treatment labels are called as a marked case set for short, and N chronic kidney disease cases without treatment labels are called as a unmarked case set for short;
step 2: training initial classifier based on feature view
Step 2.1: initializing a characteristic view: randomly extracting 1 case from the marked case set, and taking corresponding symptoms, syndrome and machine summaries and syndromes as 3 initialized feature views respectively; each characteristic view is a matrix representation formed by a plurality of characteristic vectors;
step 2.2: training an initial classifier: based on the labeled case set, training 3 different classifiers, namely a classifier based on symptom features, a classifier based on syndrome machine summary features and a classifier based on syndrome features, on 3 feature views by using an ML-KNN method, and sequentially labeling the classifiers as classifiers 1,2 and 3; after each training process is finished, updating 3 corresponding feature views, namely sequentially adding three features of symptoms, syndrome and machine summary and syndrome of the current case into the corresponding feature views until all the marked cases are trained; in the embodiment, in order to ensure that the features do not repeatedly appear, a complete word matching method is used for judging whether the same features appear in the current feature view or not in the process of adding the features;
and step 3: optimizing classifier
Step 3.1: and (4) predicting a classification result: the trained 3 classifiers are used to perform classification prediction on the unlabeled case set, as follows:
Figure BDA0002362538650000051
wherein v represents a classifier number,
Figure BDA0002362538650000052
a label representing case j predicted using classifier v,
Figure BDA0002362538650000053
represents the probability that the label of case j is t, where t is 1,2,.., K is the total number of therapeutic (labels) given a classifier v;
step 3.2: updating the prediction result: the phenomenon of class imbalance may exist in the labeled case set, that is, the number of related cases of a certain label is far less than that of related cases of another label, so that the classification accuracy is reduced; to address this issue, we use tag correlation to update their prediction results;
defining the label correlation as the correlation between every two of K kinds of labels, and calculating the following steps:
Figure BDA0002362538650000054
C(ta,tb)≠C(tb,ta)
wherein,
Figure BDA0002362538650000055
indicating simultaneous inclusion of a tag taAnd tbThe number of cases (a) of (b),
Figure BDA0002362538650000056
representing the inclusion of only the tag tbThe number of cases of (c);
the result of the prediction is updated and,
Figure BDA0002362538650000057
representing the probability of the label of case j being t after updating the classifier,
Figure BDA0002362538650000058
a vector representation representing the prediction probability over K labels;
Figure BDA0002362538650000059
Figure BDA00023625386500000510
wherein,
Figure BDA00023625386500000511
representing the probability that case j is not associated with label t, as follows:
Figure BDA00023625386500000512
Figure BDA00023625386500000513
representing the probability that case j is associated with label t, as follows:
Figure BDA00023625386500000514
step 3.3: calculating the global reliability of the prediction case j:
Figure BDA0002362538650000061
wherein Hv(j, t) represents the reliability of case j on label t as follows:
Figure BDA0002362538650000062
step 3.4: and (3) label transmission: performing descending order according to the global reliability, and selecting the first n cases as a reliable case set BvTransferring reliable case sets among the 3 classifiers, removing the reliable case sets corresponding to the unmarked case sets, adding the reliable case sets into the marked case sets to form new marked case sets, and training each classifier again on the new marked case sets by using an ML-KNN method so as to optimize the classifiers; as shown in fig. 2, the classifier 1 corresponds to a reliable case set B1The information is transmitted to classifiers 2 and 3, and so on, so that the classifier 1 is based on the reliable case set B2Reliable case set B3Training again;
step 3.5: repeating the steps 3.1-3.4 until the unmarked case set is empty; obtaining 3 final trained classifiers;
and 4, step 4: outputting the optimal classification result
Selecting an optimal prediction result by using a method combining vector splicing and weight distribution;
Figure BDA0002362538650000063
α+β+γ=1
wherein, the parameters alpha, beta and gamma are respectively the weight distribution of 3 classifiers, and the parameters are optimized and adjusted according to the experimental result; p is a radical ofjIs a K-dimensional probability vector, and each element corresponds to the probability of a label (law).
By adopting the technical scheme of the invention, the test results are shown in the following table 1.
The treatment results in table 1 are analyzed by Tibetan physicians in the Tibetan hospital of Qinghai province, and the results show that the accuracy of the label of the treatment predicted by using the method provided by the invention reaches 83.4%, which is superior to the accuracy of the treatment prediction performed by the existing method. Compared with traditional manual methods for Tibetan diagnosis prediction, the method can provide objective treatment recommendation for doctors, provides auxiliary decision for the diagnosis process of the doctors, can effectively improve the treatment accuracy rate particularly for young doctors with insufficient experience, and provides better diagnosis and treatment schemes for patients.
TABLE 1
Figure BDA0002362538650000064
Figure BDA0002362538650000071
While the invention has been described with reference to specific embodiments, any feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise; all of the disclosed features, or all of the method or process steps, may be combined in any combination, except mutually exclusive features and/or steps.

Claims (1)

1. A Tibetan medicine diagnosis assisting device based on multi-label learning, which comprises an input device for inputting symptoms, syndrome outlines and syndromes, and a classifier for receiving the input symptoms, syndrome outlines and syndromes, wherein the classifier is obtained by training the following processes:
step 1: constructing a data set comprising: m chronic kidney disease cases with treatment labels are called as a marked case set for short, and N chronic kidney disease cases without treatment labels are called as a unmarked case set for short;
step 2: training initial classifier based on feature view
Step 2.1: initializing a characteristic view: randomly extracting 1 case from the marked case set, and taking corresponding symptoms, syndrome and machine summaries and syndromes as 3 initialized feature views respectively;
step 2.2: training an initial classifier: training a classifier based on symptom features, a classifier based on syndrome machine summary features and a classifier based on syndrome features on 3 feature views by using an ML-KNN method based on a labeled case set, and sequentially labeling the classifiers as classifiers 1,2 and 3; and updating the corresponding 3 feature views after the training process is completed each time; repeating the training until all the marked cases are trained;
and step 3: optimizing classifier
Step 3.1: and (4) predicting a classification result: using 3 classifiers obtained by training to carry out classification prediction on the unmarked case set to obtain a prediction result
Figure FDA0003267214400000011
Figure FDA0003267214400000012
Wherein v represents a classifier number,
Figure FDA0003267214400000013
a probability vector representing the classifier v predicting case j,
Figure FDA0003267214400000014
representing the probability that the label of case j is t under the condition that the classifier is v, wherein t is 1, 2.
Step 3.2: updating the prediction result:
Figure FDA0003267214400000015
Figure FDA0003267214400000016
wherein,
Figure FDA0003267214400000017
representing the probability of the label of case j being t under the condition of classifier v after updating,
Figure FDA0003267214400000018
a probability vector representing the updated classifier v prediction case j;
Figure FDA0003267214400000019
represent the probability that case j is not associated with label t:
Figure FDA00032672144000000110
Figure FDA0003267214400000021
representing the probability that case j is associated with label t:
Figure FDA0003267214400000022
C(ta,tb) Indicates the label taAnd a label tbThe correlation of (a):
Figure FDA0003267214400000023
C(ta,tb)≠C(tb,ta)
wherein,
Figure FDA0003267214400000024
representing simultaneous inclusion of labels t in a marked case setaAnd tbThe number of cases (a) of (b),
Figure FDA0003267214400000025
representing only labels t in a marked case setbThe number of cases of (c);
step 3.3: calculating the global reliability of the prediction case j:
Figure FDA0003267214400000026
wherein Hv(j, t) represents the reliability of case j on label t:
Figure FDA0003267214400000027
step 3.4: performing descending order according to the global reliability, and selecting the first n cases as a reliable case set BvAnd transferring reliable case sets among the 3 classifiers, removing the reliable case sets from the corresponding unmarked case sets, adding the reliable case sets into the marked case sets to form new marked case sets, and using ML-The KNN method trains each classifier again on the new labeled case set;
step 3.5: repeating the steps 3.1-3.4 until the unmarked case set is empty; obtaining 3 final trained classifiers;
and 4, step 4: outputting an optimal classification result:
Figure FDA0003267214400000028
where α, β, γ are preset weights for classifiers 1,2, 3, respectively, and α + β + γ is 1.
CN202010026148.8A 2020-01-10 2020-01-10 Tibetan medicine diagnosis auxiliary device based on multi-label learning Active CN111128375B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010026148.8A CN111128375B (en) 2020-01-10 2020-01-10 Tibetan medicine diagnosis auxiliary device based on multi-label learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010026148.8A CN111128375B (en) 2020-01-10 2020-01-10 Tibetan medicine diagnosis auxiliary device based on multi-label learning

Publications (2)

Publication Number Publication Date
CN111128375A CN111128375A (en) 2020-05-08
CN111128375B true CN111128375B (en) 2021-11-02

Family

ID=70487697

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010026148.8A Active CN111128375B (en) 2020-01-10 2020-01-10 Tibetan medicine diagnosis auxiliary device based on multi-label learning

Country Status (1)

Country Link
CN (1) CN111128375B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111667917A (en) * 2020-06-10 2020-09-15 北京小白世纪网络科技有限公司 Method, system and equipment for realizing traditional Chinese medicine diagnosis and treatment based on neural network
CN114093445B (en) * 2021-11-18 2024-05-28 芽米科技(广州)有限公司 Patient screening marking method based on partial multi-marking learning

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107122583A (en) * 2017-03-10 2017-09-01 深圳大学 A kind of method of syndrome differentiation and device of Syndrome in TCM element
CN110164519A (en) * 2019-05-06 2019-08-23 北京工业大学 A kind of classification method for being used to handle electronic health record blended data based on many intelligence networks
CN110335684A (en) * 2019-06-14 2019-10-15 电子科技大学 The intelligent dialectical aid decision-making method of Chinese medicine based on topic model technology

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10706545B2 (en) * 2018-05-07 2020-07-07 Zebra Medical Vision Ltd. Systems and methods for analysis of anatomical images

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107122583A (en) * 2017-03-10 2017-09-01 深圳大学 A kind of method of syndrome differentiation and device of Syndrome in TCM element
CN110164519A (en) * 2019-05-06 2019-08-23 北京工业大学 A kind of classification method for being used to handle electronic health record blended data based on many intelligence networks
CN110335684A (en) * 2019-06-14 2019-10-15 电子科技大学 The intelligent dialectical aid decision-making method of Chinese medicine based on topic model technology

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Application of Multi-label Learning Model for Chronic Kidney Disease Syndrome Classification;Liwen Peng,Xiaolin Zhu,Huan Liao,Peng Zhang;《2019 IEEE 5th International Conference on Computer and Communications (ICCC)》;20191231;全文 *
Traditional Chinese Medicine (TCM) Diagnosis Model Building Based on Multi-label Classification;Lu Zhou1;Guang-geng Li2;Yu-mei Zhou1;Dan Yin1;Yan Sun1;Yan Zheng;《MATEC Web of Conferences》;20181231;第232卷;全文 *
面向慢性肾小球肾炎的中医组方辅助决策的研究;程甜甜;《中国优秀硕士学位论文全文数据库 医药卫生科技辑》;20180815(第08期);全文 *
面向慢性肾脏病中医辨证的计算机辅助决策研究;彭黎文;《万方数据库》;20190916;全文 *

Also Published As

Publication number Publication date
CN111128375A (en) 2020-05-08

Similar Documents

Publication Publication Date Title
CN109920501B (en) Electronic medical record classification method and system based on convolutional neural network and active learning
JP7522167B2 (en) Systems and methods for model-assisted cohort selection
CN109670179B (en) Medical record text named entity identification method based on iterative expansion convolutional neural network
WO2021227511A1 (en) Complication onset risk prediction method and system based on electronic medical record big data, and terminal and storage medium
JP5523342B2 (en) Method and apparatus for refining similar case search
CN110442840B (en) Sequence labeling network updating method, electronic medical record processing method and related device
Tashkandi et al. Efficient in-database patient similarity analysis for personalized medical decision support systems
JP7464800B2 (en) METHOD AND SYSTEM FOR RECOGNITION OF MEDICAL EVENTS UNDER SMALL SAMPLE WEAKLY LABELING CONDITIONS - Patent application
CN110348019B (en) Medical entity vector conversion method based on attention mechanism
CN108231146B (en) Deep learning-based medical record model construction method, system and device
CN111128375B (en) Tibetan medicine diagnosis auxiliary device based on multi-label learning
CN116910172B (en) Follow-up table generation method and system based on artificial intelligence
WO2021114635A1 (en) Patient grouping model constructing method, patient grouping method, and related device
CN112489769A (en) Intelligent traditional Chinese medicine diagnosis and medicine recommendation system for chronic diseases based on deep neural network
Chen et al. Modified bidirectional encoder representations from transformers extractive summarization model for hospital information systems based on character-level tokens (AlphaBERT): development and performance evaluation
CN112466462B (en) EMR information association and evolution method based on deep learning of image
Ningrum et al. A deep learning model to predict knee osteoarthritis based on nonimage longitudinal medical record
Gavrilov et al. Feature extraction method from electronic health records in Russia
CN114783603A (en) Multi-source graph neural network fusion-based disease risk prediction method and system
CN113838583A (en) Intelligent drug efficacy evaluation method based on machine learning and application thereof
Maheshwari et al. Quantum machine learning applied to electronic healthcare records for ischemic heart disease classification
Miao Using machine learning algorithms to predict diabetes mellitus based on Pima Indians Diabetes dataset
Zhao et al. Knowledge guided feature aggregation for the prediction of chronic obstructive pulmonary disease with Chinese EMRs
Sun et al. TSRNet: Diagnosis of COVID-19 based on self-supervised learning and hybrid ensemble model
CN109875522B (en) Method for predicting consistency of pathological scores after prostate puncture and radical operation

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