CN106169165B - Symptom hierarchy association and prediction method for diagnosis and treatment data - Google Patents
Symptom hierarchy association and prediction method for diagnosis and treatment data Download PDFInfo
- Publication number
- CN106169165B CN106169165B CN201510643963.8A CN201510643963A CN106169165B CN 106169165 B CN106169165 B CN 106169165B CN 201510643963 A CN201510643963 A CN 201510643963A CN 106169165 B CN106169165 B CN 106169165B
- Authority
- CN
- China
- Prior art keywords
- diagnosis
- treatment
- patient
- records
- symptom
- 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.)
- Expired - Fee Related
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 141
- 208000024891 symptom Diseases 0.000 title claims abstract description 59
- 238000000034 method Methods 0.000 title claims abstract description 36
- 201000010099 disease Diseases 0.000 claims abstract description 58
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 58
- 238000013507 mapping Methods 0.000 claims abstract description 22
- 238000009826 distribution Methods 0.000 claims description 37
- 238000005070 sampling Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 8
- 239000013598 vector Substances 0.000 claims description 7
- 230000008520 organization Effects 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 2
- 238000005315 distribution function Methods 0.000 claims description 2
- 238000005065 mining Methods 0.000 abstract description 6
- 238000007418 data mining Methods 0.000 abstract description 3
- 230000002265 prevention Effects 0.000 abstract description 3
- 230000002708 enhancing effect Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000010354 integration Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
Images
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention relates to data mining of unstructured diagnosis and treatment information, and discloses a diagnosis and treatment data-oriented symptom hierarchy association and prediction method. The method constructs a symptom theme hierarchical space based on a hierarchical theme model, adopts a maximum probability criterion to realize diagnosis and treatment records and hierarchical mapping of patients, and comprehensively considers multiple attribute information of diagnosis and treatment places, ages, sexes and time of the patients to realize dynamic prediction of diseases. The invention provides an integrated unified framework for automatically mining diagnosis and treatment information from three levels of diseases, diagnosis and treatment records and patients, thereby enhancing the understanding that patients and doctors have deep understanding and intuition on symptoms, influencing factors, diagnosis and treatment methods and correlation, dynamic development and evolution of the diseases in different periods, facilitating the realization of management, tracking and prediction on the diseases, the diagnosis and treatment records and the patients, and providing better information services such as early prevention, prediction, diagnosis and treatment of the diseases for the patients.
Description
Technical Field
The invention relates to data mining of unstructured diagnosis and treatment information, and discloses a diagnosis and treatment data-oriented symptom hierarchy association and prediction method.
Background
With the development of medical informatization, hospitals establish diagnosis and treatment medical records for patients and record diagnosis and treatment information of the patients in detail so as to facilitate tracking and management of the diagnosis and treatment information of the patients, a large amount of diagnosis and treatment information of the patients forms a disease diagnosis and treatment information space for many times, data mining is carried out for the diagnosis and treatment information, association between diseases and symptoms is mined from the diagnosis and treatment information, association relation between the diagnosis and treatment records and the patients is obtained, management, tracking and prediction of the diagnosis and treatment records and the patients are scientifically realized, better information service is provided for the patients, and the method is a new research topic in the field of current medical data analysis and has important theoretical and application values.
The medical record usually records the information of the doctor for the patient to make examinations and diagnoses all the time, including structured attribute information such as the time, place, age, and sex of the patient, and also includes the unstructured information described by keywords, such as the description of the patient on the disease symptoms, the treatment plan proposed by the doctor for the disease, and the like. The existing mining method for unstructured data in diagnosis and treatment records lacks a unified and integrated framework. ParikshitSondhi et al made meaningful attempts in this direction, and for unstructured text medical data, proposed a graph model-based symptom association method for discovering association relationships between planar structures of diseases, but it could not discover hierarchical associations between disease topics, and hierarchical associations could discover associations between disease topics more intuitively and vividly. In addition, the structured attributes in the diagnosis and treatment records provide rich description about diseases for unstructured diagnosis and treatment text information, for example, diagnosis and treatment time, age and location information can facilitate tracking of dynamic evolution of disease association along with time and location, the elements are fully considered in the mining process to more accurately discover the association among the diseases, the relationship among disease symptoms and influence factors thereof can be found, and the tracking, management and prediction of the diseases are facilitated.
The invention provides an effective way for discovering hierarchical association by using a hierarchical topic model, and David M.Blei and the like discover a topic hierarchical structure implied in a document abstract by using the hierarchical topic model, and provides a symptom hierarchical association and prediction method fusing multiple elements based on the hierarchical topic model aiming at the characteristic of less integration, hierarchy and dynamic consideration in the existing unstructured diagnosis and treatment record mining method, thereby comprehensively considering multiple factors, more intuitively, vividly and accurately discovering the topic hierarchical association among symptoms and realizing efficient and accurate prediction of diseases; based on the hierarchical topic structure, each diagnosis and treatment record of the patient is mapped to the corresponding node by utilizing a maximum probability mapping rule, so that hierarchical organization of diagnosis and treatment information and clustering of patient groups are realized; and finally, predicting based on the hierarchical structure, predicting the possible health crisis of the patient according to the path of the tree hierarchy where the patient is located and the node information of the corresponding diagnosis and treatment for the existing patient, mapping the existing diagnosis and treatment record of the new patient to the corresponding node, and determining the path where the new patient is located according to the maximum probability principle, thereby realizing the prediction.
Disclosure of Invention
In order to solve the characteristics of less integration, layering and dynamic consideration in the existing unstructured diagnosis and treatment record mining method, the invention provides a symptom association and prediction method based on a layered topic model, which is oriented to unstructured diagnosis and treatment data, constructs a symptom topic layered space, adopts a maximum probability criterion to realize the hierarchical mapping of diagnosis and treatment records and patients, and comprehensively considers various attribute information of the patients to realize the dynamic prediction of diseases.
The invention discloses a symptom level association and prediction method, which comprises the following steps:
step 1, constructing a diagnosis and treatment information space according to diagnosis and treatment record information of a patient;
step 4, comprehensively considering various attribute information of the patient to realize dynamic prediction of the disease;
and 5, carrying out expanded application according to the disease prediction result.
The method provided by the invention provides an integrated unified framework for automatically mining diagnosis and treatment information from three levels of diseases, diagnosis and treatment records and patients, further enhances the deep understanding and visual cognition of the patients and doctors on symptoms, influencing factors, diagnosis and treatment methods and correlation, dynamic development and evolution of the diseases in different periods, is convenient for realizing the management, tracking and prediction of the diseases, the diagnosis and treatment records and the patients, and provides better information services for the patients, such as early prevention, prediction, diagnosis and treatment of the diseases and the like.
Drawings
FIG. 1 is an overall framework of the present invention;
FIG. 2 is a schematic diagram of the spatial composition of clinical information;
FIG. 3 is a symptom topic hierarchy generation diagram;
FIG. 4 is a schematic illustration of a medical record and patient mapping;
fig. 5 is a flowchart of a symptom prediction process.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to specific embodiments and the accompanying drawings.
Fig. 1 is a flowchart of steps of a symptom level association and prediction method according to the present invention, and as shown in fig. 1, the present invention provides a symptom level association and prediction method, which includes the following steps:
s1, constructing a diagnosis and treatment information space according to the diagnosis and treatment record information of the patient;
step S2, based on the diagnosis and treatment information space of step S1, obtaining a symptom topic hierarchy space by using a hierarchy topic model, and obtaining hierarchy information of symptoms and diagnosis and treatment words in the hierarchy topic space;
step S3, according to the symptom theme hierarchical space obtained in the step S2, carrying out hierarchical mapping on the diagnosis and treatment records and the patients by using a maximum probability criterion to obtain the hierarchical information of the diagnosis and treatment records and the patients;
step S4, calculating the probability of the medical record to be diagnosed appearing in the historical medical record of the patient by comprehensively considering various attribute information of age, diagnosis time, place and sex of the diagnosis record of the patient by using the diagnosis record obtained in the step S3 and the hierarchical set information of the patient, thereby realizing the dynamic prediction of the disease;
and step S5, carrying out expanded application on the disease prediction result. The similarity correlation between the diagnosis and treatment records and the similarity correlation between the patients obtained in the step S4 can effectively realize hierarchical management of the patients and the records, facilitate deep analysis of the influence factors of the doctors and the patients on the diseases, obtain knowledge of the evolution of the diseases, facilitate more scientific tracking management of the records and the patients, and provide better information service for the patients.
The above steps are described in detail below.
S1, constructing a diagnosis and treatment information space according to the diagnosis and treatment record information of the patient;
the medical history of each patient is the diagnosis and treatment of each timeThe information collection, the diagnosis and treatment information is composed of words describing diseases, symptoms, treatment schemes and the like, and the word collection forms a diagnosis and treatment information word list CVAnd V represents the length of the word list, each diagnosis and treatment information can be represented as a vector di＝(li；agei；gi；ti；pi1,pi2,...,pin,,...,piV,)，liIndication of diagnosis and treatment diLocation information, ageiDenotes diAge of the patient at the time of presentation, giIndicates the sex of the patient, tiIndication of diagnosis and treatment diTime of (p)ijThe expression wjIn diagnosis and treatment diThe frequency of occurrence of (a). Medical record pat for each patientn＝(dn,1,dn,2,...,dn,Kn) Kn denotes patient patnThe number of diagnoses. Set of patients { pat1,pat2,...,patNAnd N represents the number of patients in the patient set, and constitutes a diagnosis and treatment information space. Fig. 2 is a schematic view of diagnosis and treatment information space consisting of disease, diagnosis and treatment records and patients.
Step S2, obtaining a symptom topic hierarchical space by using a hierarchical topic model based on the diagnosis and treatment information space of the step S1;
the symptom topic hierarchical space in step S2 is as shown in fig. 3, and based on a hierarchical topic model, the diagnosis and treatment information space is constructed into a symptom potential topic hierarchical tree by introducing a nested chinese restaurant process as a priori of a hierarchical structure/tree structure, where topics represented by each node in the hierarchical tree are represented as distributions on a symptom/treatment word list in diagnosis and treatment records, nodes at a high level represent relatively broad disease symptom topics, and nodes at a low level represent relatively narrow disease symptom topics. The step S2 specifically includes:
step S21, sampling the probability distribution of the topic on the diagnosis and treatment record word list;
for each topic k belonging to T in a topic tree structure T with the depth of L, the probability distribution β k of the sampling k on the vocabulary meets β k-Dirichlet (η), wherein the hyperparameter η controls the smoothness of the topic-diagnosis and treatment word distribution;
step S22, sampling the path from the root node to the leaf node in the tree structure for each patient;
for a set of patients { pat1,pat2,...,patNEach patient pat inn(N ∈ {1,2, 3.. eta., N }), a path c from a root node to a leaf node is sampled from the tree structurenSatisfy cn-nCRP (γ), where γ is a parameter in nCRP that controls the tree structure, nCRP (γ) is a nested chinese restaurant process with a tree structure parameter of γ;
s23, sampling distribution vectors of each diagnosis and treatment record of the patient on each layer;
for patnMedical record dn,i(i belongs to {1,2, 3.,. Kn }), and sampling diagnosis and treatment record dn,iDistribution vector theta on each layeriSatisfies thetaiI { m, pi } -GEM (m, pi) (. cndot.), the GEM (. cndot.) is a GEM distribution function when using a Stick-breaking construction method (Stick-breaking constraints), wherein the parameter m controls the balance of distribution on each subject layer, and pi determines the strictness of the obeying parameter m;
step S24, sampling a symptom word from the position represented by each symptom word in the diagnosis and treatment record;
for medical record dn,iThe position represented by each word in (a) is processed as follows:
-distribution vector θ from the last sampling stepiSampling a hierarchy z for the locationijSatisfy zij|θ～Mult(θi) Mult (·) is a polynomial distribution;
-from z already sampledij,cnFor that position, sample a word wijSatisfy wij|{zij,cn,β}～Mult(βcn[zij]) Wherein β controls the distribution of topic-words, zijHierarchy information obtained for the sampling;
wherein, the setting of 4 hyper-parameters { η, gamma, m, pi } will affect the shape and distribution of the symptom hierarchical tree theme structure, so that the expected symptom hierarchical tree structure can be obtained by adjusting the hyper-parameters.
Step S3, according to the symptom theme hierarchical space obtained in the step S2, the diagnosis and treatment records and the patients are mapped in a hierarchical mode by using the maximum probability criterion;
fig. 4 is a schematic diagram of the hierarchical mapping of the medical records and the patients in step S3, which is to map the medical records of each patient to nodes in a tree structure based on the obtained hierarchical tree of symptom potential subjects to form a hierarchical structure of the medical records of the patients, further map the patients to the nodes in the tree structure according to the mapping result of the medical records to form a hierarchical association organization and management of three levels of symptoms, medical records and patients. The step S3 specifically includes:
step S31, calculating probability distribution of the diagnosis and treatment records of the patient appearing in different levels according to the level distribution corresponding to the symptom words in the diagnosis and treatment records, and mapping the diagnosis and treatment records to the level topic nodes with the maximum probability;
probability distribution p (z) of occurrence of clinical records of patients at different levelsj|dn,i) Is calculated as follows:
wherein,
representing medical records dn,iNumber of Chinese words, wikDenotes dn,iThe k-th word in (1), p (z)j|wik,cn) Denotes dn,iWord w inikAppears on path cnMiddle hierarchy zjThe probability of (d);
step S32, according to the levels of each diagnosis and treatment record of the patient, calculating the probability distribution of the patient at different levels, and mapping the patient to the level subject node with the maximum probability;
probability distribution p (z) of patients appearing at different levelsj|patn) The calculation formula of (a) is as follows:
step S4, dynamically predicting the disease by comprehensively considering various attribute information of the patient:
fig. 5 is a flow chart showing the dynamic prediction of diseases by fusing various attribute information. The method is characterized in that a record similarity calculation method comprehensively considering a plurality of attribute information of diagnosis and treatment places, ages, sexes and time of patients is adopted. The step S4 specifically includes:
step S41, if the patient to be predicted is already in the symptom topic hierarchical tree, directly positioning the patient to the corresponding node in the hierarchical tree; if the diagnosis records do not exist in the topic hierarchical tree, according to the new diagnosis records and the symptom topic hierarchical structure obtained in the step S2, respectively mapping the diagnosis records to corresponding nodes by using a maximum topic probability mapping rule based on the method shown in the step S3;
step S42, similarity calculation is carried out on the diagnosis and treatment record of the patient to be predicted and other records on the positioned node, in the similarity calculation process, a plurality of attribute information of diagnosis and treatment place, age, sex and time of the patient and disease description and diagnosis and treatment information in the diagnosis and treatment record are comprehensively considered, and the calculation mode can effectively measure the influence of multiple factors on the disease, so that more accurate disease prediction and diagnosis and treatment are realized;
calculating dkAnd other diagnosis and treatment records d of the same nodejThe similarity calculation function of (a) is:
wherein,
weight representing probability difference of different diagnosis records in the same level, l representing diagnosis location, t representing diagnosis time, g representing sex of patient, and age representing patientAge, p (z | d)k) Representing medical records dkProbability at level z, p (z | d)j) Representing medical records djProbability at level z;
representing attributes calculating a piecewise function, skRepresenting medical records dkλ s represents a threshold corresponding to the segment of the attribute s, asRepresenting a corresponding function value when the attribute s exceeds the range;
step S43, based on the similarity of the diagnosis and treatment records obtained through calculation, obtaining the probability distribution of new diagnosis and treatment records in the historical medical records of the patient, realizing the prediction of the disease of the patient, and recommending the information of the disease treatment scheme of the patient based on the treatment information of the disease in the similar diagnosis and treatment records;
according to the mapping rule, after a subject node corresponding to the disease range of the patient is obtained, the time, the place, the sex, the age attribute of the patient and other diagnosis and treatment records corresponding to the same node and the probability distribution of the diagnosis and treatment records of other patients appearing in the node are integrated to calculate the prediction probability p (d)j|patn) Prediction of medical record djPatient pat in the futurenThe probability score of the occurrence in the medical record is calculated according to the following formula:
wherein, the patient pat is shownnThe number of levels z to which the diagnosis record is mapped, dkIndicates patnThe kth medical record at level z.
Step S5, carrying out expanding application according to the disease prediction result:
after the disease of the patient to be predicted is predicted, the patient can be prevented, treated and treated in an early stage by referring to treatment schemes of other patients with similar diseases, and structured attributes and unstructured attributes of similar diagnosis and treatment records are comprehensively analyzed, so that the deep understanding and visual understanding of the patient and doctors on symptoms, influencing factors, diagnosis and treatment methods, correlation, dynamic development and evolution of the disease in different periods can be further enhanced, the management, tracking and prediction of the disease, the diagnosis and treatment records and the patient can be conveniently realized, and better information services such as early prevention, prediction, diagnosis and treatment and the like of the disease can be provided for the patient. Meanwhile, the level correlation discovery of the three layers of diseases, diagnosis and treatment records and patients generated by the method provided by the patent also provides an effective way for a hospital to more scientifically manage medical records.
Claims (5)
1. A symptom hierarchy association and prediction method for diagnosis and treatment data comprises the following steps:
step 1, constructing a diagnosis and treatment information space according to a plurality of pieces of diagnosis and treatment record information of patients;
step 2, acquiring a symptom theme hierarchical space by using a hierarchical theme model based on the diagnosis and treatment information space;
step 3, according to the obtained symptom theme hierarchical space, carrying out hierarchical mapping on diagnosis and treatment records and patients by using a maximum probability criterion;
step 4, calculating the probability of the medical record to be diagnosed appearing in the historical medical record of the patient by comprehensively considering various attribute information of the diagnosis and treatment record information of the patient, and obtaining similarity correlation between the diagnosis and treatment records and similarity correlation between the patients;
step 5, realizing hierarchical management of patients and diagnosis and treatment records by the similarity correlation between the diagnosis and treatment records and the similarity correlation between the patients obtained in the step 4;
the diagnosis and treatment information space aims at expressing the diagnosis and treatment medical record information of the patient consisting of the diagnosis and treatment place, age, sex, time, disease symptom description and diagnosis and treatment scheme information of the patient; the step 2 is based on the diagnosis and treatment information space and the level topic model, and a symptom topic level space is constructed: constructing a diagnosis and treatment information space into a symptom potential theme hierarchical tree by introducing a nested Chinese restaurant process as a priori of a hierarchical structure/tree structure, wherein themes represented by each node in the hierarchical tree are represented as the distribution on a symptom/treatment word list in diagnosis and treatment records, nodes at a high level represent disease symptom themes at a relatively upper level, and nodes at a low level represent disease symptom themes at a relatively lower level;
the step 2 specifically comprises:
step 21, sampling probability distribution of topics on a diagnosis and treatment record word list;
step 22, sampling a path from a root node to a leaf node in a tree structure for each patient;
step 23, sampling distribution vectors of each diagnosis and treatment record of the patient on each layer;
and 24, sampling a symptom word from the position represented by each symptom word in the diagnosis and treatment record.
2. The method of claim 1, wherein step 21 satisfies β k-Dirichlet (η) for each topic k e T in the topic tree structure T of depth L, the probability distribution β k of sample k on the vocabulary, wherein the hyperparameter η controls the smoothness of the topic-diagnosis-and-treat distribution, and step 22 for the set of patients { pat1,pat2,...,patNEach patient pat inn(N ∈ {1,2, 3.. eta., N }), a path c from a root node to a leaf node is sampled from the tree structurenSatisfy cn-nCRP (γ), where γ is a parameter in nCRP that controls the tree structure, nCRP (γ) is a nested chinese restaurant process with a tree structure parameter of γ; step 23 PatnMedical record dn,i(i belongs to {1,2, 3.,. Kn }), and sampling diagnosis and treatment record dn,iDistribution vector theta on each layeriSatisfies thetaiI { m, pi } -GEM (m, pi) (. cndot.), the GEM (. cndot.) is a GEM distribution function when using a Stick-breaking construction method (Stick-breaking constraints), wherein the parameter m controls the balance of distribution on each subject layer, and pi determines the strictness of the obeying parameter m; step 24 pairs of medical records dn,The position represented by each word in the list is processed as follows;
according to the distribution vector theta obtained by the last step of samplingiSampling a hierarchy z for the locationijSatisfy zij|θ～Mult(θi) Mult (·) is a polynomial distribution;
according to z already sampledij,cnFor that position, sample a word wijSatisfy the following requirementswij|{zij,cn,β}～Mult(βcn[zij]) Wherein β controls the distribution of topic-words, zijThe setting of 4 hyper-parameters { η, gamma, m, pi } will affect the shape and distribution of the symptom hierarchical tree theme structure, so the expected symptom hierarchical tree structure can be obtained by adjusting the hyper-parameters.
3. The method according to claim 1, wherein the step 3 performs hierarchical mapping on the diagnosis and treatment records and the patients, aiming at mapping the diagnosis and treatment record of each patient to nodes in a tree structure based on the obtained hierarchical tree of symptom potential subject, forming a hierarchical structure of the diagnosis and treatment records of the patients, further mapping the patients to the nodes in the tree structure according to the mapping result of the diagnosis and treatment records, and forming hierarchical association organization and management of three levels of symptoms, diagnosis and treatment records and patients;
the step 3 specifically includes:
step 31, calculating probability distribution of the diagnosis and treatment records of the patient appearing in different levels according to the level distribution corresponding to the symptom words in the diagnosis and treatment records, and mapping the diagnosis and treatment records to level topic nodes with the maximum probability;
and 32, calculating probability distribution of the patient at different levels according to the level of each diagnosis and treatment record of the patient, and mapping the patient to the level topic node with the maximum probability.
4. The method of claim 3, wherein the patient's medical records of step 31 have a probability distribution p (z) of appearing at different levelsj|dn,i) Is calculated as follows:
step 32 probability distribution p (z) of patient appearance at different levelsj|patn) The calculation formula of (a) is as follows:
wherein Kn is patient patnThe number of the diagnosis and treatment records.
5. The method according to claim 1, wherein step 4 first locates a node corresponding to the potential topic hierarchical tree to which the patient is mapped, and performs a similar probability distribution calculation on the new record of the patient and other medical records on the node to predict the probability distribution of the new medical record appearing in the historical medical record of the patient, characterized by a record similarity calculation method comprehensively considering a plurality of attribute information of medical location, age, gender and time of the patient;
the step 4 specifically includes:
step 41, directly positioning a patient to be predicted existing in the subject hierarchical tree to a corresponding node in the hierarchical tree, and mapping the patient to be predicted, which does not exist in the subject hierarchical tree, to the corresponding node according to a new diagnosis and treatment record of the patient according to the method in the step 3;
42, carrying out similarity calculation on the diagnosis and treatment record of the patient to be predicted and other records on the positioned node, wherein in the similarity calculation process, a plurality of attribute information of diagnosis and treatment places, ages, sexes and time of the patient and disease description and diagnosis and treatment information in the diagnosis and treatment record are comprehensively considered, and the calculation mode can effectively measure the influence of multiple factors on the disease, so that more accurate disease prediction and diagnosis and treatment are realized;
and 43, obtaining the probability distribution of the new diagnosis and treatment records in the historical medical records of the patient based on the similarity of the diagnosis and treatment records obtained through calculation, realizing the prediction of the disease of the patient, and recommending the information in the aspect of the disease treatment scheme of the patient based on the treatment information of the disease in the similar diagnosis and treatment records.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201520323857 | 2015-05-20 | ||
| CN2015203238577 | 2015-05-20 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106169165A CN106169165A (en) | 2016-11-30 |
| CN106169165B true CN106169165B (en) | 2020-06-16 |
Family
ID=57358879
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510643963.8A Expired - Fee Related CN106169165B (en) | 2015-05-20 | 2015-10-09 | Symptom hierarchy association and prediction method for diagnosis and treatment data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106169165B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107122587A (en) * | 2017-03-22 | 2017-09-01 | 上海商保通健康科技有限公司 | Layer-stepping personalized health trend evaluation system based on big data |
| CN107591204B (en) * | 2017-08-01 | 2019-02-26 | 山东大学 | Rehospitalization prediction method and system based on time series portrait |
| CN109559819B (en) * | 2018-10-27 | 2024-06-18 | 平安医疗健康管理股份有限公司 | Disease type determining method and device based on sample processing and terminal |
| CN109545370B (en) * | 2018-10-27 | 2023-08-22 | 平安医疗健康管理股份有限公司 | Disease type determining method and device based on sample processing and terminal |
| CN109523404B (en) * | 2018-10-27 | 2024-06-25 | 平安医疗健康管理股份有限公司 | Data analysis-based illegal behavior detection method, device and terminal |
| CN112463973B (en) * | 2019-09-06 | 2024-07-26 | 医渡云（北京）技术有限公司 | Medical knowledge graph construction method, device, medium and electronic equipment |
| CN110853764B (en) * | 2019-11-28 | 2023-11-14 | 成都中医药大学 | A diabetes syndrome prediction system |
| CN111178444B (en) * | 2019-12-31 | 2023-06-02 | 山东中医药大学第二附属医院 | Traditional Chinese medicine formula treatment effect statistical method based on vector analysis |
| CN111445993B (en) * | 2020-03-25 | 2023-08-08 | 北京嘉和海森健康科技有限公司 | Hierarchical relationship processing method and device for diseases and electronic equipment |
| CN111462909B (en) * | 2020-03-30 | 2024-04-05 | 讯飞医疗科技股份有限公司 | Disease evolution tracking and disease condition prompting method and device and electronic equipment |
| CN112365988B (en) * | 2020-11-10 | 2023-08-04 | 杭州市肿瘤医院 | A Prognosis Prediction System |
| CN114496283A (en) * | 2022-02-15 | 2022-05-13 | 山东大学 | Disease prediction system based on path reasoning, storage medium and equipment |
| CN119008015B (en) * | 2024-10-24 | 2025-04-11 | 江苏迈鼎科技(集团)有限公司 | A system for determining venous thromboembolism prevention plans based on diagnosis and treatment information |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1839391A (en) * | 2003-06-25 | 2006-09-27 | 美国西门子医疗解决公司 | Systems and methods for automated diagnosis and decision support in breast imaging |
| CN1914617A (en) * | 2004-02-03 | 2007-02-14 | 美国西门子医疗解决公司 | Systems and methods for automated diagnosis and decision support for heart related diseases and conditions |
| CN102214232A (en) * | 2011-06-28 | 2011-10-12 | 东软集团股份有限公司 | Method and device for calculating similarity of text data |
| CN102567464A (en) * | 2011-11-29 | 2012-07-11 | 西安交通大学 | Theme map expansion based knowledge resource organizing method |
| CN102985924A (en) * | 2011-02-14 | 2013-03-20 | 松下电器产业株式会社 | Similar case retrieval device and similar case retrieval method |
| CN103020454A (en) * | 2012-12-15 | 2013-04-03 | 中国科学院深圳先进技术研究院 | Method and system for extracting morbidity key factor and early warning disease |
| CN103119444A (en) * | 2010-04-21 | 2013-05-22 | 米密德诊断学有限公司 | Markers and determinants for distinguishing bacterial from viral infections and methods of use thereof |
| CN103390058A (en) * | 2013-07-29 | 2013-11-13 | 北京理工大学 | Domain knowledge browsing method based on knowledge map |
| CN103955703A (en) * | 2014-04-25 | 2014-07-30 | 杭州电子科技大学 | Medical image disease classification method based on naive Bayes |
| CN104615894A (en) * | 2015-02-13 | 2015-05-13 | 上海中医药大学 | Traditional Chinese medicine diagnosis method and system based on k-nearest neighbor labeled specific weight characteristics |
| CN104881463A (en) * | 2015-05-22 | 2015-09-02 | 清华大学深圳研究生院 | Reference medical record search method and device based on structural medical record database |
-
2015
- 2015-10-09 CN CN201510643963.8A patent/CN106169165B/en not_active Expired - Fee Related
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1839391A (en) * | 2003-06-25 | 2006-09-27 | 美国西门子医疗解决公司 | Systems and methods for automated diagnosis and decision support in breast imaging |
| CN1914617A (en) * | 2004-02-03 | 2007-02-14 | 美国西门子医疗解决公司 | Systems and methods for automated diagnosis and decision support for heart related diseases and conditions |
| CN103119444A (en) * | 2010-04-21 | 2013-05-22 | 米密德诊断学有限公司 | Markers and determinants for distinguishing bacterial from viral infections and methods of use thereof |
| CN102985924A (en) * | 2011-02-14 | 2013-03-20 | 松下电器产业株式会社 | Similar case retrieval device and similar case retrieval method |
| CN102214232A (en) * | 2011-06-28 | 2011-10-12 | 东软集团股份有限公司 | Method and device for calculating similarity of text data |
| CN102567464A (en) * | 2011-11-29 | 2012-07-11 | 西安交通大学 | Theme map expansion based knowledge resource organizing method |
| CN103020454A (en) * | 2012-12-15 | 2013-04-03 | 中国科学院深圳先进技术研究院 | Method and system for extracting morbidity key factor and early warning disease |
| CN103390058A (en) * | 2013-07-29 | 2013-11-13 | 北京理工大学 | Domain knowledge browsing method based on knowledge map |
| CN103955703A (en) * | 2014-04-25 | 2014-07-30 | 杭州电子科技大学 | Medical image disease classification method based on naive Bayes |
| CN104615894A (en) * | 2015-02-13 | 2015-05-13 | 上海中医药大学 | Traditional Chinese medicine diagnosis method and system based on k-nearest neighbor labeled specific weight characteristics |
| CN104881463A (en) * | 2015-05-22 | 2015-09-02 | 清华大学深圳研究生院 | Reference medical record search method and device based on structural medical record database |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106169165A (en) | 2016-11-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106169165B (en) | Symptom hierarchy association and prediction method for diagnosis and treatment data | |
| Senan et al. | Score and Correlation Coefficient‐Based Feature Selection for Predicting Heart Failure Diagnosis by Using Machine Learning Algorithms | |
| WO2023202508A1 (en) | Cognitive graph-based general practice patient personalized diagnosis and treatment scheme recommendation system | |
| Shukla et al. | A literature review in health informatics using data mining techniques | |
| Ma et al. | Attention-guided deep graph neural network for longitudinal Alzheimer’s disease analysis | |
| Dyagilev et al. | Learning (predictive) risk scores in the presence of censoring due to interventions | |
| Alzahani et al. | An overview of data mining techniques applied for heart disease diagnosis and prediction | |
| Carmona et al. | Evolutionary fuzzy rule extraction for subgroup discovery in a psychiatric emergency department | |
| Montani et al. | Improving structural medical process comparison by exploiting domain knowledge and mined information | |
| US20250149172A1 (en) | System for forecasting a mental state of a subject and method | |
| Nistal-Nuño | Machine learning applied to a Cardiac Surgery Recovery Unit and to a Coronary Care Unit for mortality prediction | |
| Adi et al. | Stroke risk prediction model using machine learning | |
| Colella et al. | Studying variables affecting the length of stay in patients with lower limb fractures by means of Machine Learning | |
| Das et al. | Artificial intelligent reliable doctor (AIRDr.): Prospect of disease prediction using reliability | |
| Patil et al. | An Adaptive parameter free data mining approach for healthcare application | |
| Loku et al. | Automated medical data analyses of diseases using big data | |
| Revathy et al. | GUI based heart disease prediction using deep neural networks | |
| Chitra et al. | Effective heart attack prediction method using machine learning algorithm | |
| Prabhavathi et al. | Diagnosis of cardiac disease utilizing machine learning techniques and dense neural networks | |
| Jiang et al. | PercolationDF: A percolation-based medical diagnosis framework | |
| Parry et al. | Fuzzy ontologies for cardiovascular risk prediction-A research approach | |
| Zhang et al. | Specificity-aware federated graph learning for brain disorder analysis with functional MRI | |
| Ogundokun et al. | Review of cardiovascular disease prediction based on machine learning algorithms | |
| Kumar et al. | A hybrid approach for medical images classification and segmentation to reduce complexity: A. Kumar et al. | |
| Zhuang et al. | Continuous Identification of Sepsis-Associated Acute Heart Failure Patients: An Integrated LSTM-Based Algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200616 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |