CN107887022B - SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes - Google Patents

SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes Download PDF

Info

Publication number
CN107887022B
CN107887022B CN201711096765.XA CN201711096765A CN107887022B CN 107887022 B CN107887022 B CN 107887022B CN 201711096765 A CN201711096765 A CN 201711096765A CN 107887022 B CN107887022 B CN 107887022B
Authority
CN
China
Prior art keywords
syndrome
symptom
document
symptoms
sstm
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
CN201711096765.XA
Other languages
Chinese (zh)
Other versions
CN107887022A (en
Inventor
马甲林
陈伯伦
张琳
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.)
Beijing Derong Smart International Technology Center (L.P.)
Jiangxi Liangben Information Technology Co ltd
Original Assignee
Huaiyin Institute of Technology
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 Huaiyin Institute of Technology filed Critical Huaiyin Institute of Technology
Priority to CN201711096765.XA priority Critical patent/CN107887022B/en
Publication of CN107887022A publication Critical patent/CN107887022A/en
Application granted granted Critical
Publication of CN107887022B publication Critical patent/CN107887022B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
  • Medical Preparation Storing Or Oral Administration Devices (AREA)

Abstract

The invention provides a SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes, which comprises the steps of firstly inputting an electronic patient document diagnosis and treatment record data set which can be identified by a computer program; secondly, a symptom-syndrome topic model (SSTM) is constructed, parameter solution is carried out, and the SSTM is trained; and then inputting the prediction samples, and performing incremental training by adopting loose SSTM to perform intelligent syndrome diagnosis. The method can be applied to a traditional Chinese medicine learning system, greatly reduces the difficulty in learning and mastering the traditional Chinese medicine diagnosis knowledge, can also be applied to various traditional Chinese medicine intelligent diagnosis and treatment systems, and has important significance for promoting the intellectualization and standardization of syndrome diagnosis.

Description

SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes
Technical Field
The invention relates to an intelligent diagnosis method for traditional Chinese medicine, in particular to an intelligent diagnosis method for traditional Chinese medicine syndromes based on SSTM.
Background
The diagnosis and treatment based on syndrome differentiation is one of the characteristics of the traditional Chinese medicine, and a key link in the diagnosis and treatment process of the traditional Chinese medicine is syndrome differentiation, namely: doctors need to make syndrome diagnosis by the information of the patients' four diagnostic methods-inspection and inquiry, so as to determine the treatment scheme and the prescription compatibility, and the syndrome differentiation reflects the knowledge, experience and level of traditional Chinese medicine. Because syndromes have the characteristics of complexity, diversity, ambiguity and the like, it often takes years or even decades for traditional Chinese medicine learners, especially young doctors to master the 'syndrome differentiation' rule, and from the perspective of data mining, the rich experience and knowledge about syndrome diagnosis of traditional Chinese medicine, especially the famous and old traditional Chinese medicine, are hidden in a large amount of diagnosis and treatment history records. The existing data analysis and mining technology about traditional Chinese medicine intelligent diagnosis mainly adopts multivariate statistics and conventional data mining methods, cannot effectively deal with complex syndrome diagnosis rule mining, and cannot reach the level of practical application.
Disclosure of Invention
The purpose of the invention is as follows: the SSTM-based intelligent diagnosis method for the traditional Chinese medicine syndromes can be applied to a traditional Chinese medicine learning system, greatly reduces the difficulty in learning and mastering the traditional Chinese medicine diagnosis knowledge, can also be applied to various traditional Chinese medicine intelligent diagnosis and treatment systems, and has important significance for promoting the intellectualization and standardization of syndrome diagnosis.
The technical scheme is as follows: the SSTM-based intelligent diagnosis method for the traditional Chinese medicine syndromes comprises the following steps of:
(1) inputting an electronic patient file diagnosis and treatment record data set which can be identified by a computer program;
(2) constructing an SSTM, and solving parameters in the model;
(3) training the SSTM and storing a training result;
(4) inputting a prediction sample;
(5) intelligent syndrome diagnosis-incremental training with relaxed SSTM;
(6) outputting the diagnosis result and the symptom rule thereof;
where relaxed SSTM is an explicit constraint for SSTM to cancel random assignment of syndromes during the initialization and sampling phases.
Each piece of data of the data set is regarded as a document d and consists of one or more syndromes and a plurality of corresponding symptoms.
The step (2) comprises the following steps:
(21) SSTM is constructed, the model theme is served by a syndrome label from an explicit variable in a data set, the theme is distributed only aiming at the main symptoms, the secondary symptoms are not distributed to the theme, and one symptom sdnThe generation probability formula of (1) is as follows:
Figure BDA0001462405860000021
wherein d represents a Chinese medicine diagnosis and treatment record, s represents symptoms, and sdnRepresents the nth symptom of the document d, z represents the syndrome, zdnRepresents the syndrome to which the nth symptom of the document d belongs, y is a main symptom and a secondary symptom mark, ydn0 denotes that the nth symptom of the document d is a main symptom, ydn1 indicates that the nth symptom of the document d is a secondary symptom;
(22) SSTM parameter sampling solution is carried out, and the concrete formula is as follows:
Figure BDA0001462405860000022
wherein "-" indicates that the current position symptom t is excluded;
Figure BDA0001462405860000023
indicates the count of all the symptoms of which y is 1 after excluding the current position symptom t in the document d,
Figure BDA0001462405860000024
indicates the count of all symptoms after excluding the current position symptom t in the document d,
Figure BDA0001462405860000025
indicating that after the current position symptom t is excluded from the training set, all counts of t occurring,
Figure BDA0001462405860000026
the method comprises the steps of representing the counting of all symptoms after eliminating symptoms t at the current position in a training set, wherein V represents the number of symptoms in the training set, V is Beta (V) distribution hyper-parameter, and eta is Dirichlet (eta) distribution hyper-parameter;
Figure BDA0001462405860000027
wherein "-" indicates that the current position symptom t is excluded,
Figure BDA0001462405860000028
indicates the count of all the symptoms for which y is 0 after excluding the current position symptom t in the document d,
Figure BDA0001462405860000029
indicates the count of all symptoms after excluding the current position symptom t in the document d,
Figure BDA00014624058600000210
the counting of all occurrences of t in the syndrome with the number of k after the symptom t at the current position is eliminated in the training set is shown;
Figure BDA00014624058600000211
the counting of all symptoms in the syndrome with the number of k after the current position symptom t is eliminated in the training set is shown;
Figure BDA00014624058600000212
indicates the count of all symptoms marked as syndrome number k after excluding the current position symptom t in the document d,
Figure BDA00014624058600000213
a count representing all the symptoms marked as y ═ 0 after the current position symptom t is excluded in the document d, MdExpressing the number of syndromes in the document d, wherein v is Beta (v) distribution hyper-parameter, and alpha and Beta are Dirichlet distribution hyper-parameter;
Figure BDA0001462405860000031
for the rule parameters of syndrome k, the probability value that a certain symptom t belongs to syndrome k is calculated by adopting a mean parameter estimation method:
Figure BDA0001462405860000032
Figure BDA0001462405860000033
indicates the probability that the symptom t belongs to the syndrome with the number k;
θdand (3) representing probability parameters of all syndromes contained in the document d, specifically, adopting a mean parameter estimation method when the probability that a certain syndrome k belongs to the document d:
Figure BDA0001462405860000034
Figure BDA0001462405860000035
representing the probability that syndrome k belongs to document d.
The step (3) comprises the following steps:
(31) inputting sampling iteration times Iter, over parameters v, eta, alpha and beta and training a sample set;
(32) random initialization 1: traversing the training data set, randomly assigning a y value to each symptom s of each document in the training data set, and storing the y value by using the vector y;
(33) random initialization 2: traversing the training dataset, for each document, all symptoms s assigned y ═ 0, from Λ of that documentdRandomly distributing a syndrome number in the set, and storing the syndrome number by a vector z;
(34) re-traversing the training data set, and for each s, determining the primary and secondary symptom indicating variable y of ss1 is resampling y according to equation (2)sValue when ysRe-sampling the syndrome number of s according to a formula (3) and updating in y and z data;
(35) repeating the sampling process Iter times of the data set until Gibbs sampling is converged;
(36) the syndrome rule phi is obtained by statistics according to the formula (4)k；
(37) Output y and z, syndrome rule phik。
The loose SSTM in the step (5) is a dominant constraint condition lambda for canceling random allocation of syndromes by the SSTM in the initialization and sampling stagesd。
The step (5) comprises the following steps:
(51) inputting sampling iteration number Iter, over parameters v, eta, alpha and beta, a sample to be predicted, and y and z obtained in the step (36);
(52) random initialization 1: traversing the sample to be predicted, randomly giving a y value to each symptom s, and storing by using an update vector y;
(53) random initialization 2: traversing the sample to be predicted, and randomly distributing a syndrome number from all syndrome number sets appearing in z for all symptoms s endowed with y ═ 0;
(54) repeating the prediction samples, wherein for each s, the primary and secondary symptom indicator y of ss1 is the weight according to the formula (2)New sample ysA value; when y issRe-sampling the syndrome number of s according to a formula (3) and updating in y and z;
(55) repeating the sampling process Iter times of the data set until Gibbs sampling is converged;
(56) obtaining the syndrome diagnosis result theta of the patient to be predicted according to the formula (5) statisticsd；
(57) Outputting the syndrome diagnosis result theta of the sample to be predictedd。
Has the advantages that: compared with the prior art, the invention has the beneficial effects that: 1. a new symptom-syndrome topic model (SSTM) is proposed; 2. the SSTM model distinguishes different positions of the primary symptoms and the secondary symptoms, and reduces the interference of the secondary symptoms on syndrome diagnosis; 3. the no-syndrome label data of the patient to be diagnosed adopts loose SSTM to carry out intelligent dialectic, and the intelligent dialectic accuracy of the traditional Chinese medicine can be improved.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a graph showing the relationship between syndrome and symptoms.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings, and fig. 1 is a flow chart of the present invention.
1. Selection of training sample set and prediction sample
The specific sample data is from a national population and health science data sharing platform, 21 traditional Chinese medicine disease special subject database data are collected, 1741 diseases are related, 127,541 data are total, 51,144 data are obtained after data without syndrome marks are removed, 8,751 data are obtained after symptoms and the number of syndromes is less than 3; randomly taking 90% of the data as a training set, and taking the other 10% as a test set. The specific format of the sample data is shown in table 1.
TABLE 1 training sample data example
Figure BDA0001462405860000041
Figure BDA0001462405860000051
2. The SSTM is constructed, the model theme is served by a syndrome label from an explicit variable in a data set, the theme is distributed only aiming at the main symptoms, the secondary symptoms are not distributed, the depth relation of the symptoms and symptoms is adopted as described in figure 2, the description reflects the dialectical principle of the main symptoms and the secondary symptoms of the traditional Chinese medicine, and the traditional Chinese medicine syndrome diagnosis process can be more accurately described.
3. Parameter setting
SSTM model training phase parameters: training the SSTM according to the training algorithm, wherein the sampling iteration number Iter is 1000; the hyperparameters ν ═ 0.5, γ ═ 0.01, η ═ 0.01, α ═ K/50(K is the number of non-repetitive syndromes appearing in the training dataset), and β ═ 0.01.
Relaxation of SSTM incremental training phase parameters: training the relaxed SSTM according to the training algorithm, wherein the sampling iteration number Iter is 1000; the hyperparameter ν ═ 0.5, η ═ 0.01, α ═ K/50, and β ═ 0.01.
4. Training SSTM models
The SSTM training algorithm is as follows:
inputting: sampling iteration times Iter, over parameters v, eta, alpha and beta, and training a sample set;
and (3) outputting: vector y and z, syndrome rule
Figure BDA0001462405860000052
(1) Random initialization 1: traversing the training data set, randomly assigning a y value (y is 0, or y is 1) to each symptom s of each document in the training data set, and storing the y value by using a vector y;
(2) random initialization 2: traversing the training dataset, for each document, all symptoms s assigned y ═ 0, from Λ of that documentdRandomly assigned a syndrome number, stored by a vector z (where syndromes of all documents in the training dataset appear are numbered, Λ)dA number set of the syndromes in the document d);
(3) re-traversing the training data set, and for each s, determining the primary and secondary symptom indicating variable y of ss1 is resampling y according to equation (2)sA value; when y iss0 re-sample the syndrome number of s according to equation (3) (the adopted procedure is Λ)dConstraints) and updated in the y and z data;
(4) repeating the Sampling process (3) Iter times on the data set until Gibbs Sampling converges;
(5) the syndrome rule phi is obtained by statistics according to the formula (4)k；
(6) Outputs y and z; the syndrome rule is phik。
5. Syndrome rule based on SSTM training result
Based on the syndrome rule of the SSTM training result, the first 5 symptoms with the highest probability values are selected as shown in Table 2:
TABLE 2 example of syndrome rules
Figure BDA0001462405860000061
6. Intelligent syndrome diagnosis-relaxed SSTM for incremental training
In the intelligent syndrome diagnosis stage, the received input is a plurality of symptom information and unknown syndromes of the patient, and the SSTM model is supervised learning and cannot be directly used for incremental training of syndrome diagnosis; therefore, the invention adopts the relaxed SSTM to carry out increment training, wherein the relaxed SSTM cancels the dominant constraint condition lambda randomly distributed to syndromes in the initialization and sampling stages of the SSTMdThe relaxed SSTM can be used for carrying out incremental model training again on the symptom information of the patient to be diagnosed with the unknown syndrome label, so that an intelligent diagnosis result is obtained.
The relaxed SSTM training algorithm is as follows:
inputting: sampling iteration number Iter, over parameters v, eta, alpha and beta, a sample to be predicted, and y and z obtained in a training stage;
and (3) outputting: predicting sample syndrome diagnostic result thetad；
(1) Random initialization 1: traversing samples to be predicted, randomly assigning a y value (y is 0, or y is 1) to each symptom s, and updating a vector y;
(2) random initialization 2: traversing the sample to be predicted, and randomly distributing a syndrome number from all syndrome number sets appearing in z for all symptoms s endowed with y ═ 0;
(3) repeating the prediction samples, wherein for each s, the primary and secondary symptom indicator y of ss1 is resampling y according to equation (2)sA value; when y issRe-sampling the syndrome number of s (the syndrome number randomly distributed by the process is from all the syndrome number sets appearing in z) according to the formula (3) and updating in y and z;
(4) repeating the Sampling process (3) Iter times on the data set until Gibbs Sampling converges;
(5) obtaining the syndrome diagnosis result theta of the patient to be predicted according to the formula (5) statisticsd。
(6) Outputting the syndrome diagnosis result theta of the sample to be predictedd。
7. Example of diagnostic results
Output thetadThe first m syndromes with the highest probability in the syndrome set are the diagnosis results of the patient to be diagnosed, and the mark vectors y of the main symptoms and the minor symptoms of all the symptoms of the patient are outputdOutput rules for storing K syndromes
Figure BDA0001462405860000071
Inputting symptoms of a patient to be predicted: cough, dyspnea, chest distress, yellow and sticky sputum, fever, nasal discharge, pharyngalgia, thirst, yellow urine, dry and hard stool, red tongue, yellow coating, slippery and rapid pulse, cough, weak voice, clear and thin sputum, dyspnea, spontaneous perspiration, anorexia, abdominal distension, mental fatigue, hypodynamia, loose stool, pale tongue, white coating, soft and slow pulse, dyspnea and cough.
And outputting a diagnosis result:
syndrome diagnosis: syndrome of phlegm-heat accumulating in the lung, spleen-lung qi deficiency, lung-kidney yin deficiency,
principal symptoms (inevitable symptoms): cough and asthma, chest distress, fever, watery nasal discharge, pharyngalgia, thirst, constipation, red tongue, yellow tongue coating, slippery rapid pulse, clear and thin sputum, dyspnea, asthenia, pale tongue, and cough and asthma;
secondary (probable) symptoms: yellow phlegm, sticky pus, drowned yellow, cough, weakness, spontaneous perspiration, anorexia, abdominal distention, mental fatigue, loose stool, white fur, and soft-superficial and slow pulse.
The syndrome diagnosis problem related to the method can be classified as a multi-label classification problem of machine learning: one sample consists of a plurality of symptoms and one or more syndromes, the symptoms can be regarded as content data of the sample, and the syndromes can be regarded as classification labels of the sample. Therefore, the evaluation indexes of the multi-label classification task can be used for comparing and evaluating the method results, wherein the evaluation indexes of the accuracy and the comprehensive evaluation index F1 are selected for evaluating the results, and the method (SSTM) provided by the invention is compared with two classification algorithms of a Support Vector Machine (SVM) and Naive Bayes (NB), as shown in Table 3:
TABLE 3 comparison of syndrome diagnosis results by three methods
Method Accuracy (%) F1 value
NB 68.21 60.21
SVM 75.65 67.98
SSTM 79.14 72.61
As can be seen from Table 3, the method has the highest accuracy and the highest comprehensive index value for the intelligent diagnosis of the syndrome of traditional Chinese medicine.

Claims (2)

1. An SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes is characterized by comprising the following steps:
(1) inputting an electronic patient file diagnosis and treatment record data set which can be identified by a computer program;
(2) constructing a symptom-syndrome theme model SSTM, and solving parameters in the model;
(3) training the SSTM and storing a training result;
(4) inputting a prediction sample;
(5) intelligent syndrome diagnosis-incremental training with relaxed SSTM;
(6) outputting the diagnosis result and the symptom rule thereof;
wherein, the relaxed SSTM is an explicit constraint condition that the SSTM cancels the random distribution of the syndromes in the initialization and sampling stages;
the step (2) comprises the following steps:
(21) SSTM is constructed, the model theme is served by a syndrome label from an explicit variable in a data set, the theme is distributed only aiming at the main symptoms, the secondary symptoms are not distributed to the theme, and one symptom sdnThe generation probability formula of (1) is as follows:
Figure FDA0003249490560000011
wherein d represents a Chinese medicine diagnosis and treatment record, s represents symptoms, and sdnRepresents the nth symptom of the document d, z represents the syndrome, zdnRepresents the syndrome to which the nth symptom of the document d belongs, y is a main symptom and a secondary symptom mark, ydn0 denotes that the nth symptom of the document d is a main symptom, ydn1 indicates that the nth symptom of the document d is a secondary symptom;
(22) SSTM parameter sampling solution is carried out, and the concrete formula is as follows:
Figure FDA0003249490560000012
wherein "-" indicates that the current position symptom t is excluded;
Figure FDA0003249490560000013
indicates the count of all the symptoms of which y is 1 after excluding the current position symptom t in the document d,
Figure FDA0003249490560000014
indicates the count of all symptoms after excluding the current position symptom t in the document d,
Figure FDA0003249490560000015
indicating that after the current position symptom t is excluded from the training set, all counts of t occurring,
Figure FDA0003249490560000016
the method comprises the steps of representing the counting of all symptoms after eliminating symptoms t at the current position in a training set, wherein V represents the number of symptoms in the training set, V is Beta (V) distribution hyper-parameter, and eta is Dirichlet (eta) distribution hyper-parameter;
Figure FDA0003249490560000017
wherein "-" indicates that the current position symptom t is excluded,
Figure FDA0003249490560000021
indicates the count of all the symptoms for which y is 0 after excluding the current position symptom t in the document d,
Figure FDA0003249490560000022
indicates the count of all symptoms after excluding the current position symptom t in the document d,
Figure FDA0003249490560000023
the counting of all occurrences of t in the syndrome with the number of k after the symptom t at the current position is eliminated in the training set is shown;
Figure FDA0003249490560000024
the counting of all symptoms in the syndrome with the number of k after the current position symptom t is eliminated in the training set is shown;
Figure FDA0003249490560000025
indicates the count of all symptoms marked as syndrome number k after excluding the current position symptom t in the document d,
Figure FDA0003249490560000026
a count representing all the symptoms marked as y ═ 0 after the current position symptom t is excluded in the document d, MdExpressing the number of syndromes in the document d, wherein v is Beta (v) distribution hyper-parameter, and alpha and Beta are Dirichlet distribution hyper-parameter;
Figure FDA0003249490560000027
for the rule parameters of syndrome k, the probability value that a certain symptom t belongs to syndrome k is calculated by adopting a mean parameter estimation method:
Figure FDA0003249490560000028
Figure FDA0003249490560000029
indicates the probability that the symptom t belongs to the syndrome with the number k;
θdand (3) representing probability parameters of all syndromes contained in the document d, specifically, adopting a mean parameter estimation method when the probability that a certain syndrome k belongs to the document d:
Figure FDA00032494905600000210
Figure FDA00032494905600000211
representing the probability that syndrome k belongs to document d;
the step (3) comprises the following steps:
(31) inputting sampling iteration number Iter, over parameters v, eta, alpha and beta, and training a sample set;
(32) random initialization 1: traversing the training sample set, randomly giving a y value to each symptom s of each document in the training sample set, and storing the y value by using the vector y;
(33) random initialization 2: traversing the training sample set, all symptoms s given y ═ 0 for each document, from Λ of that documentdRandomly assigning a syndrome number, storing by vector z, ΛdThe number sets of all syndromes in the document d are set;
(34) re-traversing the training sample set, and for each s, when the primary and secondary symptom indicating variables y of ssResampling y according to equation (2) when 1sValue when ysResampling the syndrome number of s according to a formula (3) and updating in the vector y and z data as 0;
(35) repeating the sampling process Iter times of the sample set until Gibbs sampling is converged;
(36) the syndrome rule is obtained by statistics according to the formula (4)
Figure FDA0003249490560000031
(37) Output vectors y and z, syndrome rules
Figure FDA0003249490560000032
The step (5) comprises the following steps:
(51) inputting sampling iteration number Iter, over parameters v, eta, alpha and beta, a sample to be predicted, and vectors y and z obtained in the step (37);
(52) random initialization 1: traversing a sample to be predicted, randomly giving a y value to each symptom s, and storing by using an update vector y;
(53) random initialization 2: traversing samples to be predicted, and randomly distributing a syndrome number from all syndrome number sets appearing in the vector z for all symptoms s endowed with y ═ 0;
(54) re-traversing the sample to be predicted, and for each s, determining the primary and secondary symptom indicator variable y of ssResampling y according to equation (2) when 1sA value; when y issResampling the syndrome number of s according to formula (3) and updating in vectors y and z for 0;
(55) repeating the sampling process Iter times of the samples to be predicted until Gibbs sampling is converged;
(56) obtaining the syndrome diagnosis result theta of the patient to be predicted according to the formula (5) statisticsd；
(57) Outputting the syndrome diagnosis result theta of the sample to be predictedd。
2. The SSTM-based intelligent diagnosis method of syndromes in traditional Chinese medicine according to claim 1, wherein each piece of data in the data set in step (1) is regarded as a document d consisting of one or more syndromes and corresponding symptoms.
CN201711096765.XA 2017-11-09 2017-11-09 SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes Active CN107887022B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711096765.XA CN107887022B (en) 2017-11-09 2017-11-09 SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711096765.XA CN107887022B (en) 2017-11-09 2017-11-09 SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes

Publications (2)

Publication Number Publication Date
CN107887022A CN107887022A (en) 2018-04-06
CN107887022B true CN107887022B (en) 2021-10-29

Family

ID=61779638

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711096765.XA Active CN107887022B (en) 2017-11-09 2017-11-09 SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes

Country Status (1)

Country Link
CN (1) CN107887022B (en)

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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
CN110853764B (en) * 2019-11-28 2023-11-14 成都中医药大学 Diabetes syndrome prediction system
CN110970129B (en) * 2019-12-23 2022-08-16 河南中医药大学 Method for judging traditional Chinese medicine syndrome based on improved Bayesian statistics
CN111477295B (en) * 2020-04-10 2022-06-03 电子科技大学 Traditional Chinese medicine formula recommendation method and system based on latent semantic model
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
CN111834012A (en) * 2020-07-14 2020-10-27 中国中医科学院中医药信息研究所 Traditional Chinese medicine syndrome diagnosis method and device based on deep learning and attention mechanism
CN112382393A (en) * 2020-10-15 2021-02-19 天津中医药大学 PSD traditional Chinese medicine syndrome quantitative diagnosis model construction method
CN112509672A (en) * 2020-12-14 2021-03-16 河南经方云科技有限公司 Traditional Chinese medicine dialectical treatment auxiliary system and method based on artificial intelligence
CN113140310B (en) * 2021-05-06 2024-02-27 王耘 Intelligent diagnosis and treatment system for traditional Chinese medicine and creation method
CN115630307B (en) * 2022-11-29 2023-03-21 中国中医科学院中医药信息研究所 Syndrome diagnosis method and device based on cross-view angle and attention convergence

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106709520A (en) * 2016-12-23 2017-05-24 浙江大学 Topic model based medical record classification method

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106709520A (en) * 2016-12-23 2017-05-24 浙江大学 Topic model based medical record classification method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"A Biterm Topic Model for Short Texts";Xiaohui Yan等;《Proceedings of the 22nd international conference on World Wide Web》;20130531;全文 *
"Discovering Syndrome Regularities in Traditional Chinese Medicine Clinical by Topic Model";Jialin Ma等;《advances on P2P,Parallel,Grid,Cloud and Internet Computing》;20161022;第157-162页 *

Also Published As

Publication number Publication date
CN107887022A (en) 2018-04-06

Similar Documents

Publication Publication Date Title
CN107887022B (en) SSTM-based intelligent diagnosis method for traditional Chinese medicine syndromes
Houthooft et al. Predictive modelling of survival and length of stay in critically ill patients using sequential organ failure scores
Patil et al. Hybrid prediction model for type-2 diabetic patients
CN106339593B (en) Kawasaki disease classification prediction method based on medical data modeling
CN112712118A (en) Medical text data oriented filtering method and system
CN106934235A (en) Patient's similarity measurement migratory system between a kind of disease areas based on transfer learning
US20230222311A1 (en) Generating machine learning models using genetic data
CN103443809B (en) Discrimination model facility for study, methods and procedures
JP7041614B6 (en) Multi-level architecture for pattern recognition in biometric data
Pokharel et al. Temporal tree representation for similarity computation between medical patients
CN113593708A (en) Sepsis prognosis prediction method based on integrated learning algorithm
CN116564409A (en) Machine learning-based identification method for sequencing data of transcriptome of metastatic breast cancer
Karrar The effect of using data pre-processing by imputations in handling missing values
Özkan et al. Effect of data preprocessing on ensemble learning for classification in disease diagnosis
Prasad et al. Modeling risk prediction of diabetes—A preventive measure
Jain et al. An improved supervised classification algorithm in healthcare diagnostics for predicting opioid habit disorder
Khanna et al. An Integrated Approach towards the prediction of Likelihood of Diabetes
Siddiqa et al. Robust Length of Stay Prediction Model for Indoor Patients.
CN116432856A (en) Pipeline dynamic early warning method and device based on CNN-GLSTM model
CN111144424A (en) Personnel feature detection and analysis method based on clustering algorithm
Yun et al. Forecasting of heart rate variability using wrist-worn heart rate monitor based on hidden Markov model
Chang et al. Non‐homogeneous continuous‐time Markov chain with covariates: Applications to ambulatory hypertension monitoring
CN110265151B (en) Learning method based on heterogeneous temporal data in EHR
CN114420300B (en) Chinese senile cognitive impairment prediction model
Mistry et al. Estimating missing data and determining the confidence of the estimate data

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
EE01 Entry into force of recordation of patent licensing contract

Application publication date: 20180406

Assignee: Huai'an Huairui Information Technology Co.,Ltd.

Assignor: HUAIYIN INSTITUTE OF TECHNOLOGY

Contract record no.: X2022980025546

Denomination of invention: An Intelligent Diagnosis Method of TCM Syndrome Based on SSTM

Granted publication date: 20211029

License type: Common License

Record date: 20221209

EE01 Entry into force of recordation of patent licensing contract
TR01 Transfer of patent right

Effective date of registration: 20230911

Address after: Room 1274, Block D, Building 2, Changcheng Road, Yanqing Garden, Zhongguancun, Yanqing District, Beijing, 102100

Patentee after: Beijing Derong Smart International Technology Center (L.P.)

Address before: Room 602, Unit D, Building B3, Fenglu Xintiandi, Nanchang Economic and Technological Development Zone, Jiangxi 330052

Patentee before: Jiangxi Liangben Information Technology Co.,Ltd.

Effective date of registration: 20230911

Address after: Room 602, Unit D, Building B3, Fenglu Xintiandi, Nanchang Economic and Technological Development Zone, Jiangxi 330052

Patentee after: Jiangxi Liangben Information Technology Co.,Ltd.

Address before: 223003 Jiangsu Huaian economic and Technological Development Zone, 1 East Road.

Patentee before: HUAIYIN INSTITUTE OF TECHNOLOGY

TR01 Transfer of patent right