CN110111884B - CMKMC-based man-machine cooperative intelligent medical aid decision-making system - Google Patents
CMKMC-based man-machine cooperative intelligent medical aid decision-making system Download PDFInfo
- Publication number
- CN110111884B CN110111884B CN201910359377.9A CN201910359377A CN110111884B CN 110111884 B CN110111884 B CN 110111884B CN 201910359377 A CN201910359377 A CN 201910359377A CN 110111884 B CN110111884 B CN 110111884B
- Authority
- CN
- China
- Prior art keywords
- diagnosis
- clinical
- module
- disease
- medical
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT 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
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H80/00—ICT specially adapted for facilitating communication between medical practitioners or patients, e.g. for collaborative diagnosis, therapy or health monitoring
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention relates to a CMKMC-based man-machine cooperative intelligent medical aid decision-making system, which comprises: the disease initial diagnosis module is used for matching the clinical data with the disease node attributes of the dynamic medical cognitive attribute knowledge base so as to determine the suspected disease range of the clinical disease, provide clinical examination guidance and obtain the initial diagnosis result; the accurate diagnosis module is used for carrying out quantitative disease risk probability analysis on suspected diseases by using the clinical data which cannot be diagnosed accurately by the disease initial diagnosis module through the machine learning diagnosis network cluster to obtain an accurate diagnosis result; the comprehensive diagnosis module is used for verifying the accurate diagnosis result of the machine learning network cluster aiming at clinical difficult and complicated diseases and the initial diagnosis node information in the medical cognitive attribute knowledge base, and comprehensively evaluating to obtain a diagnosis conclusion; and the self-evolution module is used for analyzing and relearning the misdiagnosed cases of the system, updating the reasoning path and the node attribute of the medical cognitive attribute knowledge base and adjusting the internal parameters of the machine learning diagnosis network cluster. The invention has the characteristics of high diagnosis accuracy and self-evolution.
Description
Technical Field
The invention belongs to the technical field of artificial intelligence and intelligent medical treatment, and particularly relates to a CMKMC-based man-machine cooperative intelligent medical treatment auxiliary decision-making system, which integrates a dynamic medical treatment cognitive attribute knowledge base and a machine learning diagnosis network cluster, and assists clinical decision-making based on a man-machine cooperative consultation mechanism.
Background
Current clinical medical decisions are largely determined by medical practitioners in conjunction with their own experience and patient-related examination reports, the results of which are often highly correlated with the medical practitioner's own level. With the continuous improvement of the requirements of the modern society on the quality of medical services, the requirements on the quality and quantity of medical practitioners are also continuously improved. The culture speed of qualified professional medical practitioners is far from keeping up with the requirement of the construction of the current medical system, so that the quantity and quality of a large number of medical practitioners are insufficient.
According to the statistics of the national health administration, by 2018, almost four billion people exist in China, the number of medical practitioners is less than three hundred and more than ten thousand, and the number of experienced high-level physicians is extremely small; and only less than 10% of hospitals in China are regarded as high-level medical institutions, but more than half of patients in China need to be treated. The shortage and imbalance of high-quality medical resources are very serious problems faced by the Chinese medical industry. Practitioners in a healthcare facility are heavily loaded and statistics show that the time from patient interview to diagnosis by an outpatient at a large healthcare facility is typically no more than fifteen minutes. The medical quality is also reduced by the drastic increase in the working pressure of the doctors and the significant decrease in the time for diagnostic thinking.
An intelligent medical aid decision-making technology supported by an AI technology is one of the emerging research hotspots in recent years, and the essence of the technology is to apply artificial intelligence and a machine learning technology to summarize and summarize medical big data characteristics and generate a corresponding diagnosis network for clinical disease diagnosis of a current patient so as to assist a doctor in clinical decision-making. The intelligent medical technology can balance medical resources while improving the working efficiency of doctors, so that higher-quality intelligent diagnosis and treatment service can be obtained in regions with laggard medical conditions.
The intelligent medical aid decision-making system has important significance for improving the diagnosis efficiency of clinical diseases. On one hand, with the help of the artificial intelligent auxiliary diagnosis network, the intelligent medical auxiliary decision-making system saves the disease diagnosis time of the patient, so that a doctor can make more accurate medical decision in a shorter time. On the other hand, the intelligent medical aid decision-making system is established on the basis of a large amount of clinical data analysis, so that the state of a patient can be more comprehensively evaluated, and the risk of misdiagnosis and missed diagnosis is reduced. Therefore, the intelligent medical aid decision can reduce the workload of doctors, improve the medical efficiency and relieve the condition of insufficient clinical medical resources of most medical institutions in China.
However, the current intelligent medical assisted clinical decision making system has the following problems: due to the black box characteristic of the AI network, most of the current systems can only make a binary conclusion that the disease is ' yes ' or not ', and cannot provide a specific diagnosis process. Therefore, the doctor can not make a judgment on whether the auxiliary diagnosis basis of the system is reasonable, and the clinical practical value is reduced. Secondly, the system is not closely combined with clinical medicine cognition, so that whether the diagnosis conclusion of the machine learning network is correct or not cannot be automatically verified. The system excessively depends on the auxiliary diagnosis function of the machine learning network, and a hierarchical diagnosis mechanism is not established for clinical diseases, so that the diagnosable disease range of the system in the actual clinical process is limited.
Disclosure of Invention
Based on the defects in the prior art, the invention provides a CMKMC-based man-machine cooperative intelligent medical aid decision-making system, which integrates a dynamic medical cognitive attribute knowledge base and a machine learning diagnosis network cluster, and assists clinical decision based on a man-machine cooperative consultation mechanism.
In order to achieve the purpose, the invention adopts the following technical scheme:
a CMKMC-based man-machine cooperative intelligent medical aid decision-making system comprises:
the disease initial diagnosis module is used for matching the clinical data with the disease node attributes of the dynamic medical cognitive attribute knowledge base so as to determine the suspected disease range of the clinical disease, provide clinical examination guidance and obtain the initial diagnosis result;
the accurate diagnosis module is used for carrying out quantitative disease risk probability analysis on suspected diseases by the clinical data which cannot be diagnosed by the disease initial diagnosis module through the machine learning diagnosis network cluster to obtain an accurate diagnosis result;
the comprehensive diagnosis module is used for verifying the accurate diagnosis result of the machine learning network cluster aiming at clinical difficult and complicated diseases and the initial diagnosis node information in the medical cognitive attribute knowledge base, and comprehensively evaluating to obtain a diagnosis conclusion;
and the self-evolution module is used for analyzing and relearning the misdiagnosed cases of the system, updating the reasoning path and the node attribute of the medical cognitive attribute knowledge base and adjusting the internal parameters of the machine learning diagnosis network cluster.
As a preferred scheme, the dynamic medical cognitive attribute knowledge base in the disease initial diagnosis module comprises a plurality of clinical disease node entities, node attributes and cognitive inference paths among nodes; wherein, the clinical disease node entity consists of a clinical expression sub-node entity and a clinical examination sub-node entity; the clinical manifestation sub-node entity consists of a plurality of groups of clinical symptom manifestations, and the clinical examination sub-node entity consists of a plurality of clinical examination means; the cognitive inference path between the nodes is based on the current clinical medical cognition, is based on the analysis and knowledge summarization induction of clinical medical big data, and establishes a logical inference relation by taking key clinical characteristics as judgment characteristics to customize the clinical medical cognition into a plurality of cognitive inference paths connecting all the nodes.
Preferably, the accurate diagnosis module comprises a data completion layer, a cluster diagnosis layer and a conclusion fusion layer;
the data completion layer is used for completing missing data of the clinical examination data by adopting mathematical statistics, regression analysis and data fitting;
the cluster diagnosis layer is used for performing respective independent diagnosis analysis on the supplemented clinical data through a plurality of mutually independent basic machine learning diagnosis networks;
and the conclusion fusion layer is used for carrying out comprehensive decision on the result output by the cluster diagnosis layer to obtain an accurate diagnosis result.
Preferably, the plurality of mutually independent basic machine learning diagnosis networks in the cluster diagnosis layer comprise SVC, RBF-NN, ANFIS and Naive Bayes.
As a preferred scheme, the conclusion fusion layer comprises an integrated analysis layer and a decision return layer, wherein the integrated analysis layer is used for performing comprehensive decision on the result output by the cluster diagnosis layer by adopting a plurality of integrated learners and finally providing an estimation of the prevalence probability; and the decision returning layer is used for returning three disease diagnosis results with the highest risk probability and corresponding risk probability values for the system.
Preferably, the multiple ensemble learners in the ensemble analysis layer include Adaboost, Logistic regression, random forest, ID3 decision tree, and Bagger weighted average.
Preferably, the comprehensive diagnosis module comprises a checking analysis module, a diagnosis analysis module and a clinical examination guidance module;
the checking and analyzing module is used for checking and analyzing the disease doubtful range of the primary diagnosis result and several diseases corresponding to which the disease risk is higher than a preset threshold in the fine diagnosis result;
the diagnosis analysis module is used for outputting the corresponding clinical performance attribute analysis result and the inference path in the disease node with or without the coincidence of the system initial diagnosis node and the precise diagnosis result together with the initial diagnosis result and the precise diagnosis result so as to assist the decision of doctors;
and the clinical examination guide module is used for increasing the required clinical examination feedback to the doctor.
As a preferred scheme, the self-evolution module comprises a misdiagnosis case analysis module, a misdiagnosis sample relearning module, a feature extraction module and an unsupervised training analysis module;
the misdiagnosis case analysis module is used for analyzing the attributes and characteristics of the misdiagnosis samples after the number of the misdiagnosis samples is accumulated to a set threshold value;
the misdiagnosis sample relearning module is used for carrying out secondary learning on misdiagnosis samples and updating the diagnosis probability expectation of the machine learning network, so that the machine learning network cluster is evolved;
the characteristic extraction module is used for extracting main characteristics of diseases from similar cases;
the unsupervised cluster analysis module is used for searching potential relations among the disease nodes and establishing a cognitive inference path; and when the relation between disease nodes or node attributes which are not established with the relation exceeds a threshold value and is correspondingly confirmed, updating the node attributes and the logic paths in the dynamic cognitive attribute knowledge base.
Compared with the prior art, the invention has the beneficial effects that:
(1) the establishment and evolution of the dynamic medical cognitive attribute knowledge base can quickly reduce the suspected range of diseases, make clinical examination guidance, and provide clear diagnosis and analysis paths, so that doctors are helped to improve their own medical level, the clinical diagnosis and treatment efficiency for general diseases is improved, and the diagnosis time of patients is shortened.
(2) The machine learning diagnosis network cluster is based on clinical big data analysis, and can accurately quantify the disease risk probability for reference of clinicians. The diagnosis accuracy is further improved, and meanwhile, the problem of difficulty in differential diagnosis of difficult and complicated diseases is solved more effectively.
(3) The CMKMC-based man-machine cooperative intelligent medical aid decision-making system integrates a dynamic medical cognitive attribute knowledge base and a machine learning diagnosis network cluster, takes a man-machine cooperative consultation mechanism as a leading comprehensive diagnosis, and realizes mutual verification, mutual supervision and mutual learning of the three, so that the reliability, accuracy and timeliness of a diagnosis result are improved to the greatest extent, and the clinical use value of the system is remarkably enhanced.
(4) In clinical disease experiments, the CMKMC intelligent medical disease diagnosis system is proved to have high diagnosis accuracy, can give more accurate diagnosis results for different clinical diseases of a plurality of departments, has the overall diagnosis accuracy of 95.27 percent, has higher diagnosis sensitivity for each disease, and is not easy to cause missed diagnosis.
Drawings
FIG. 1 is a schematic structural diagram of a CMKMC-based human-computer cooperative intelligent medical aid decision-making system according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a dynamic cognitive attribute knowledge base structure according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a precise diagnosis module of a machine learning network cluster according to an embodiment of the present invention;
FIG. 4 is a flow chart of the human-computer cooperative diagnosis of the CMKMC based human-computer cooperative intelligent medical assistance decision making system according to the embodiment of the present invention;
FIG. 5 shows the clinical disease diagnosis result of the CMKMC-based human-computer cooperative intelligent medical aid decision-making system in the experiment according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
The invention is described in further detail below with reference to the attached drawing figures:
fig. 1 is a schematic structural diagram of a CMKMC-based human-computer cooperative intelligent medical assistance decision-making system according to an embodiment of the present invention.
The CMKMC-based man-machine cooperative intelligent medical aid decision-making system comprises a dynamic medical cognitive attribute knowledge base disease initial diagnosis module, a machine learning network cluster accurate diagnosis module, a comprehensive diagnosis module and a self-evolution module. The CMKMC-based man-machine cooperative intelligent medical aid decision-making system disclosed by the embodiment of the invention simultaneously uses the preliminary diagnosis of a medical dynamic cognitive attribute knowledge base and the accurate diagnosis of a machine learning network cluster, and mutually assists in logic and function. The disease initial diagnosis module rapidly reduces the suspected range of clinical diseases while solving simple disease diagnosis, and recommends currently required clinical examination to doctors; the machine learning diagnosis network cluster is used for carrying out quantitative disease probability analysis on diseases which are difficult to identify and diagnose clinically from the perspective of medical big data; the clinical practicability and the diagnosis accuracy of the system are improved.
And the disease initial diagnosis module of the dynamic medical cognitive attribute knowledge base is used for matching clinical data (namely patient EMR data) with the disease node attributes of the dynamic medical cognitive attribute knowledge base so as to determine the suspected disease range of clinical diseases, provide clinical examination guidance and obtain the initial diagnosis result. Based on the knowledge base of the dynamic medical cognitive attributes, the suspected range of clinical diseases is rapidly narrowed, and guidance is provided for follow-up required examination of patients. The dynamic medical cognitive attribute knowledge base in the disease initial diagnosis module comprises a plurality of clinical disease node entities, node attributes and cognitive inference paths among nodes; wherein, the clinical disease node entity consists of a clinical expression sub-node entity and a clinical examination sub-node entity; the clinical manifestation sub-node entity consists of a plurality of groups of clinical symptom manifestations, and the clinical examination sub-node entity consists of a plurality of clinical examination means; the cognitive inference path between the nodes is based on the current clinical medical cognition, is based on the analysis and knowledge summarization induction of clinical medical big data, and establishes a logical inference relation by taking key clinical characteristics as judgment characteristics to customize the clinical medical cognition into a plurality of cognitive inference paths connecting all the nodes. Specifically, as shown in fig. 2, the structure diagram of the dynamic cognitive attribute knowledge base according to the embodiment of the present invention is shown, the dynamic medical cognitive attribute knowledge base is established according to the refining of clinical medical knowledge and the analysis of medical big data, and the structure composition of the dynamic medical cognitive attribute knowledge base includes: 1) medical node Entity (Entity), 2) medical node Attribute (Attribute)3), cognitive inference path (cognitve logic relationship) between medical nodes. The medical node entities are divided into disease ID node entities (main nodes), and clinical symptom expression nodes and clinical examination means nodes (sub-nodes) under the administration of the disease ID node entities. Each node contains a plurality of medical attributes and logic relations among the nodes. The establishment of the cognitive inference path among the nodes is based on the analysis of clinical medical big data and the summary and induction of knowledge, and the clinical medical knowledge is abstracted into a piece of cognitive inference path by taking key clinical attributes as judgment features according to logical relations.
And the machine learning network cluster accurate diagnosis module is used for carrying out suspected disease quantitative risk probability analysis on the clinical data which cannot be diagnosed by the disease initial diagnosis module through the machine learning diagnosis network cluster so as to obtain an accurate diagnosis result. The machine learning network cluster is used for analyzing medical big data, and disease accurate diagnosis results are quantitatively provided for a medical decision system in a disease probability mode. Specifically, as shown in fig. 3, the structure diagram of the accurate diagnosis module of the machine learning network cluster according to the embodiment of the present invention is mainly divided into three layers: a data completion layer, a cluster diagnosis layer and a conclusion fusion layer;
the data completion layer is used for completing missing data of the clinical examination data by adopting mathematical statistics, regression analysis and data fitting; specifically, the data completion layer includes a plurality of modules such as mathematical statistics, regression analysis, data fitting, and the like, and estimates and recovers missing data (i.e., missing data of clinical examination data) required in the current cluster diagnosis.
The cluster diagnosis layer is used for performing respective independent diagnosis analysis on the supplemented clinical data through a plurality of mutually independent basic machine learning diagnosis networks; specifically, the cluster diagnosis layer is composed of a plurality of machine learning networks trained in test set data, wherein the machine learning networks comprise a plurality of network structures such as a support vector machine SVC, a Naive Bayes, an RBF-NN and an ANFIS. When cluster diagnosis is carried out, the system selects a plurality of proper networks from the network according to the performance of the network in the corresponding disease test set data, and adds the networks into a cluster diagnosis queue, and obtains respective independent diagnosis results according to the current patient clinical data.
And the conclusion fusion layer is used for carrying out comprehensive decision on the result output by the cluster diagnosis layer to obtain an accurate diagnosis result. Specifically, the conclusion fusion layer comprises an integrated analysis layer and a decision return layer, wherein the integrated analysis layer comprises a plurality of integrated learners including Adaboost, Logistic regression, random forest, ID3 decision tree, and bag weighted average. And calling a proper ensemble learning device to carry out secondary ensemble learning on the result made by the cluster diagnosis layer according to the diagnosis accuracy rate of the independent machine learning network in the cluster diagnosis layer in the test data set, and providing the quantized ill probability output after ensemble learning for the system. And the decision returning layer is used for carrying out comprehensive decision on the result output by the integrated analysis layer and returning a disease diagnosis result with the disease probability risk exceeding a set threshold value and a corresponding risk probability value for the system.
And the comprehensive diagnosis module is used for verifying the accurate diagnosis result of the machine learning network cluster aiming at the clinical difficult and complicated diseases and the initial diagnosis node information in the medical cognitive attribute knowledge base, and comprehensively evaluating to obtain a diagnosis conclusion. Specifically, the comprehensive diagnosis module comprises a checking analysis module, a diagnosis analysis module and a clinical examination guidance module; the checking and analyzing module is used for checking and analyzing the disease doubtful range of the primary diagnosis result and several diseases corresponding to which the disease risk is higher than a preset threshold in the fine diagnosis result; the diagnosis analysis module is used for outputting the corresponding clinical performance attribute analysis result and the reasoning path in the disease node with or without the coincidence of the system initial diagnosis node and the precision diagnosis result together with the initial diagnosis result and the precision diagnosis result to assist the decision of the doctor, and the clinical examination guidance module is used for increasing the required clinical examination feedback to the doctor.
And the self-evolution module is used for analyzing and relearning the misdiagnosed cases of the system, updating the reasoning path and the node attribute of the medical cognitive attribute knowledge base and adjusting the internal parameters of the machine learning diagnosis network cluster. Namely, the self-evolution module realizes the relearning of the system misdiagnosis medical record. After the diagnosis of the current case is finished, the machine learning diagnosis network cluster relearns the case sample with the wrong diagnosis, and improves the internal network structure and the node weight. And the dynamic cognitive attribute knowledge base rechecks the node relation in the error sample through unsupervised cluster learning, and corrects the node cognitive inference path after confirmation is obtained. Specifically, the self-evolution module comprises: the system comprises a misdiagnosis case analysis module, a machine learning network cluster misdiagnosis sample relearning module, an EMR (electromagnetic resonance) feature extraction module and an unsupervised training analysis module; after the number of misdiagnosed samples is accumulated to a set threshold value, the self-evolution module analyzes the attributes and the characteristics of the misdiagnosed samples through the misdiagnosed case analysis module, performs secondary learning on the misdiagnosed samples through the machine learning network cluster misdiagnosed sample relearning module, and updates the diagnosis probability expectation of the machine learning network, so that the machine learning network cluster accurate diagnosis module network is evolved. Meanwhile, through the EMR feature extraction module and the unsupervised cluster analysis module, the main features of diseases are extracted from similar cases, a cognitive inference path is built, and after confirmation of a clinician, node attributes and logic paths in the dynamic cognitive attribute knowledge base are updated.
As shown in fig. 4, the human-computer cooperative diagnosis process of the CMKMC-based human-computer cooperative intelligent medical assistance decision system of the present embodiment includes:
when the method is used for assisting clinical diagnosis decision-making, aiming at the current clinical examination data, a disease node is activated by a disease initial diagnosis module of a dynamic medical cognitive attribute knowledge base according to the current clinical symptoms of a patient and an examination result, disease inference diagnosis is carried out by using a logical inference and mode matching mode, and a disease suspicion range and a clinical examination suggestion are recommended according to the activation state of a network node. When serious or difficult diseases are encountered, the dynamic medical cognitive attribute knowledge base disease initial diagnosis module makes clinical examination guidance and summarizes examination data according to needs, and sends a differential diagnosis request to the machine learning network cluster accurate diagnosis module. And provides it with the following information: 1) disease area to be differentially diagnosed 2) current complete clinical examination data of the patient.
After receiving the medical data of the patient and the consultation request, the machine learning network cluster accurate diagnosis module finally feeds back the diseases with higher morbidity probability and the respective morbidity probability risks to the clinician through a plurality of layers of machine learning network diagnosis and cluster diagnosis mechanisms. And meanwhile, making positive/negative diagnosis judgment on the disease according to a set probability risk threshold.
After the accurate diagnosis process of the machine learning network cluster is completed, the comprehensive diagnosis module is responsible for automatically analyzing and organically combining the initial diagnosis result from the disease initial diagnosis module of the dynamic medical cognitive attribute knowledge base, the clinical cognitive inference path, the quantitative disease diagnosis probability result from the accurate differential diagnosis module of the machine learning network cluster and the diagnosis result of a clinician. The checking and analyzing module is used for checking and analyzing the cognitive reasoning paths of the diseases with higher disease probability and the initial diseases in the accurate diagnosis result. In the analysis process, the module inquires the reasoning path of the dynamic medical cognitive attribute knowledge base in the initial diagnosis module again according to the diagnosis result of the machine learning network cluster accurate diagnosis module, and meanwhile, the module is verified with the current clinical data of the patient. The diagnosis analysis module outputs the corresponding clinical performance attribute analysis result and the inference path in the disease node with or without the coincidence of the system initial diagnosis node and the precise diagnosis result together with the initial diagnosis result and the precise diagnosis result so as to assist the decision of doctors. The examination guiding module searches the knowledge base of the preliminary diagnosis module, and determines whether to suggest a doctor to adopt a clinical examination means with higher cost or stronger invasiveness for final diagnosis according to the accurate diagnosis result of the current system and the probability value of the disease risk. Finally, the integrated diagnostic module will provide the clinician with the following information: the machine learning network cluster accurate diagnosis module diagnoses results and the probability of illness risks. And secondly, specific clinical characteristics which accord with or object to the diagnosis result are in the system dynamic cognition attribute knowledge base. And recommending advanced clinical examination to the patient.
After the current case is diagnosed, when the number of misdiagnosed samples is accumulated to a set threshold value, the self-evolution module performs self-evolution training on the CMKMC. The method comprises the steps of firstly analyzing attributes and characteristics of misdiagnosed samples through a misdiagnosed case analysis module, carrying out secondary learning on the misdiagnosed samples through a machine learning network cluster misdiagnosed sample relearning module, adjusting confidence coefficient parameters and node values in the network, and updating diagnosis probability expectation of a machine learning network, so that a machine learning network cluster accurate diagnosis module is evolved. Meanwhile, through the EMR feature extraction module and the unsupervised training analysis module, the misdiagnosed cases are learned, the main features of similar disease cases are extracted, potential relations among the nodes are searched, and when the nodes without the relations established before or the relations among the node attributes exceed a threshold value and are confirmed by a clinician, the node attributes and the logic paths in the dynamic cognitive attribute knowledge base are updated.
TABLE 1
| Parameter name | Computing method | Meaning of parameters |
| Rate of accuracy | Number of correctly diagnosed persons/total number of cases | Coincidence of diagnostic result and real result of characterization system |
| Sensitive TP | Diagnosis patient/actual patient | Characterizing the ability of a system to detect a patient |
| Specific TN | Number of diagnosed/actual patients without disease | Characterization system ability to identify non-patient |
| Misdiagnosis rate FP | Number of false positives/number of actual non-patients | Characterizing likelihood of system non-patient diagnostic error |
| Missed diagnosis rate FN | False negative/actual number of patients | Characterizing likelihood of system patient diagnosis error |
TABLE 2
| Disease ID | The accuracy rate% | Sensitivity TP% | Specific TN% | Misdiagnosis rate FP% | The rate of missed diagnosis TN% |
| Cardiac arrhythmia | 93.98 | 94.47 | 88.23 | 11.76 | 5.52 |
| Pyelonephritis | 96.3 | 96.48 | 91.67 | 8.33 | 3.51 |
| Adenocarcinoma of lung | 93.45 | 94.61 | 79.31 | 20.68 | 5.38 |
| Hyperthyroidism | 97.36 | 98.29 | 92.3 | 7.69 | 1.7 |
As can be seen from fig. 5, table 1 and table 2, the human-computer cooperative intelligent medical aid decision-making system based on CMKMC according to the embodiment of the present invention has high diagnosis accuracy in clinical disease experiments, can provide relatively accurate diagnosis results for different clinical diseases of multiple departments, has a high overall diagnosis accuracy of 95.27%, has high diagnosis sensitivity for each disease, and is not easy to miss diagnosis.
Therefore, the CMKMC-based man-machine cooperation intelligent medical aid decision-making system has the characteristics of man-machine cooperation, high diagnosis accuracy, strong clinical practicability and self-evolution.
The CMKMC is a short term for a dynamic medical Cognitive attribute knowledge base and a Machine learning network Cluster (Cognitive medical knowledge-learning Cluster).
The foregoing has outlined rather broadly the preferred embodiments and principles of the present invention and it will be appreciated that those skilled in the art may devise variations of the present invention that are within the spirit and scope of the appended claims.
Claims (6)
1. A CMKMC-based man-machine cooperative intelligent medical aid decision-making system is characterized by comprising:
the disease initial diagnosis module is used for matching the clinical data with the disease node attributes of the dynamic medical cognitive attribute knowledge base so as to determine the suspected disease range of the clinical disease, provide clinical examination guidance and obtain the initial diagnosis result;
the accurate diagnosis module is used for carrying out quantitative disease risk probability analysis on suspected diseases by the clinical data which cannot be diagnosed by the disease initial diagnosis module through the machine learning diagnosis network cluster to obtain an accurate diagnosis result;
the comprehensive diagnosis module is used for verifying the accurate diagnosis result of the machine learning network cluster aiming at clinical difficult and complicated diseases and the initial diagnosis node information in the medical cognitive attribute knowledge base, and comprehensively evaluating to obtain a diagnosis conclusion;
the self-evolution module is used for analyzing and relearning misdiagnosed cases of the system, updating reasoning paths and node attributes of the medical cognitive attribute knowledge base and adjusting internal parameters of the machine learning diagnosis network cluster;
the dynamic medical cognitive attribute knowledge base in the disease initial diagnosis module comprises a plurality of clinical disease node entities, node attributes and cognitive inference paths among nodes; wherein, the clinical disease node entity consists of a clinical expression sub-node entity and a clinical examination sub-node entity; the clinical manifestation sub-node entity consists of a plurality of groups of clinical symptom manifestations, and the clinical examination sub-node entity consists of a plurality of clinical examination means; the cognitive inference path between the nodes is based on the current clinical medical cognition, is based on the analysis and knowledge summary induction of clinical medical big data, and establishes a logical inference relation by taking key clinical characteristics as judgment characteristics to customize the clinical medical cognition into a plurality of cognitive inference paths connecting all the nodes;
wherein the comprehensive diagnosis module comprises a checking analysis module, a diagnosis analysis module and a clinical examination guidance module;
the checking and analyzing module is used for checking and analyzing the disease doubtful range of the primary diagnosis result and several diseases corresponding to which the disease risk is higher than a preset threshold in the fine diagnosis result;
the diagnosis analysis module is used for outputting the corresponding clinical performance attribute analysis result and the inference path in the disease node with or without the coincidence of the system initial diagnosis node and the precise diagnosis result together with the initial diagnosis result and the precise diagnosis result so as to assist the decision of doctors;
the clinical examination guiding module is used for increasing the required clinical examination feedback to the doctor;
the CMKMC is short for a dynamic medical cognitive attribute knowledge base and a machine learning network cluster.
2. The CMKMC based man-machine cooperative intelligent medical aid decision making system of claim 1, wherein the precise diagnosis module comprises a data complement layer, a cluster diagnosis layer and a conclusion fusion layer;
the data completion layer is used for completing missing data of the clinical examination data by adopting mathematical statistics, regression analysis and data fitting;
the cluster diagnosis layer is used for performing respective independent diagnosis analysis on the supplemented clinical data through a plurality of mutually independent basic machine learning diagnosis networks;
and the conclusion fusion layer is used for carrying out comprehensive decision on the result output by the cluster diagnosis layer to obtain an accurate diagnosis result.
3. The CMKMC-based human-computer cooperative intelligent medical aid decision-making system of claim 2, wherein the plurality of mutually independent underlying machine learning diagnosis networks in the cluster diagnosis layer comprise SVC, RBF-NN, ANFIS, Naive Bayes.
4. The CMKMC-based man-machine cooperative intelligent medical aid decision-making system as claimed in claim 2, wherein the conclusion fusion layer comprises an integrated analysis layer and a decision return layer, the integrated analysis layer is used for performing comprehensive decision-making on the results output by the cluster diagnosis layer by using a plurality of integrated learners and finally providing an estimation of the probability of illness; and the decision returning layer is used for returning three disease diagnosis results with the highest risk probability and corresponding risk probability values for the system.
5. The CMKMC-based man-machine cooperative intelligent medical aid decision-making system as claimed in claim 4, wherein the plurality of ensemble learners in the ensemble analysis layer comprise Adaboost, Logistic regression, random forest, ID3 decision tree, and Bagger weighted average.
6. The CMKMC-based man-machine cooperative intelligent medical aid decision-making system of claim 1, wherein the self-evolution module comprises a misdiagnosis case analysis module, a misdiagnosis sample relearning module, a feature extraction module and an unsupervised training analysis module;
the misdiagnosis case analysis module is used for analyzing the attributes and characteristics of the misdiagnosis samples after the number of the misdiagnosis samples is accumulated to a set threshold value;
the misdiagnosis sample relearning module is used for carrying out secondary learning on misdiagnosis samples and updating the diagnosis probability expectation of the machine learning network, so that the machine learning network cluster is evolved;
the characteristic extraction module is used for extracting main characteristics of diseases from similar cases;
the unsupervised training analysis module is used for searching potential relations among disease nodes and establishing a cognitive inference path; and when the relation between disease nodes or node attributes which are not established with the relation exceeds a threshold value and is correspondingly confirmed, updating the node attributes and the logic paths in the dynamic cognitive attribute knowledge base.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910359377.9A CN110111884B (en) | 2019-04-30 | 2019-04-30 | CMKMC-based man-machine cooperative intelligent medical aid decision-making system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910359377.9A CN110111884B (en) | 2019-04-30 | 2019-04-30 | CMKMC-based man-machine cooperative intelligent medical aid decision-making system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110111884A CN110111884A (en) | 2019-08-09 |
| CN110111884B true CN110111884B (en) | 2021-08-24 |
Family
ID=67487681
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910359377.9A Active CN110111884B (en) | 2019-04-30 | 2019-04-30 | CMKMC-based man-machine cooperative intelligent medical aid decision-making system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110111884B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111292852A (en) * | 2020-03-12 | 2020-06-16 | 陕西领创精准医疗科技有限公司 | Encephalitis and meningitis intelligent auxiliary diagnosis system based on random forest algorithm |
| CN111383760A (en) * | 2020-04-10 | 2020-07-07 | 韩琳 | Method for establishing medical intelligent diagnosis system for nervous system diseases |
| CN112002414B (en) * | 2020-08-23 | 2024-01-26 | 吾征智能技术(北京)有限公司 | Gastric juice-based system, gastric juice-based equipment and gastric juice-based storage medium |
| CN112669967B (en) * | 2020-12-24 | 2022-09-16 | 福建福寿康宁科技有限公司 | Active health medical decision-making assisting method and equipment |
| CN113421653B (en) * | 2021-06-23 | 2022-09-09 | 平安科技（深圳）有限公司 | Medical information pushing method and device, storage medium and computer equipment |
| CN113407841B (en) * | 2021-06-25 | 2023-04-28 | 陈亮 | Method and system for automatically recommending AI (automatic indicator) scheme based on performance analysis of structured report |
| CN113838571B (en) * | 2021-09-03 | 2023-12-22 | 北京约来健康科技有限公司 | Big health medical system based on big data |
| CN114743647A (en) * | 2022-03-24 | 2022-07-12 | 康键信息技术（深圳）有限公司 | Medical data processing method, device, equipment and storage medium |
| CN115497616B (en) * | 2022-10-25 | 2023-05-23 | 杭州杏林信息科技有限公司 | Method, system, equipment and storage medium for auxiliary decision-making of infectious diseases |
| CN115757801B (en) * | 2022-12-22 | 2023-11-03 | 北京壹永科技有限公司 | Decision tree-based model training method and device for medical text |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355035A (en) * | 2016-09-28 | 2017-01-25 | 湖南老码信息科技有限责任公司 | Pneumonia prediction method and prediction system based on incremental neural network model |
| CN107491630A (en) * | 2016-06-10 | 2017-12-19 | 韩国电子通信研究院 | Clinical decision support integrated system and use its clinical decision support method |
| CN108182262A (en) * | 2018-01-04 | 2018-06-19 | 华侨大学 | Intelligent Answer System construction method and system based on deep learning and knowledge mapping |
| CN108231189A (en) * | 2017-12-12 | 2018-06-29 | 华南师范大学 | Data processing method and medical diagnosis on disease device based on dual-depth nerve learning network |
-
2019
- 2019-04-30 CN CN201910359377.9A patent/CN110111884B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107491630A (en) * | 2016-06-10 | 2017-12-19 | 韩国电子通信研究院 | Clinical decision support integrated system and use its clinical decision support method |
| CN106355035A (en) * | 2016-09-28 | 2017-01-25 | 湖南老码信息科技有限责任公司 | Pneumonia prediction method and prediction system based on incremental neural network model |
| CN108231189A (en) * | 2017-12-12 | 2018-06-29 | 华南师范大学 | Data processing method and medical diagnosis on disease device based on dual-depth nerve learning network |
| CN108182262A (en) * | 2018-01-04 | 2018-06-19 | 华侨大学 | Intelligent Answer System construction method and system based on deep learning and knowledge mapping |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110111884A (en) | 2019-08-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110111884B (en) | CMKMC-based man-machine cooperative intelligent medical aid decision-making system | |
| Sun et al. | Diagnosis and analysis of diabetic retinopathy based on electronic health records | |
| Bozkurt et al. | Using automatically extracted information from mammography reports for decision-support | |
| Duggal et al. | Prediction of thyroid disorders using advanced machine learning techniques | |
| Thirugnanam et al. | Improving the prediction rate of diabetes diagnosis using fuzzy, neural network, case based (FNC) approach | |
| CN112151170A (en) | Method for calculating a score of a medical advice for use as a medical decision support | |
| CN114639479A (en) | Intelligent diagnosis auxiliary system based on medical knowledge map | |
| Mekruksavanich | Medical expert system based ontology for diabetes disease diagnosis | |
| Tiwari et al. | A knowledge infused context driven dialogue agent for disease diagnosis using hierarchical reinforcement learning | |
| CN116110570A (en) | Diabetes auxiliary diagnosis system, text processing method and map construction method | |
| CN109213871A (en) | Patient information knowledge mapping construction method, readable storage medium storing program for executing and terminal | |
| CN111951965B (en) | Panoramic health dynamic monitoring and predicting system based on time sequence knowledge graph | |
| Pal et al. | Deep learning techniques for prediction and diagnosis of diabetes mellitus | |
| Bhat et al. | Predictions of diabetes and diet recommendation system for diabetic patients using machine learning techniques | |
| CN110085314A (en) | Intelligent analysis method, system and the equipment of medical test data | |
| CN116910172B (en) | Follow-up table generation method and system based on artificial intelligence | |
| CN109147954A (en) | The patient information processing unit of knowledge based map | |
| Huang et al. | A structure-related fine-grained deep learning system with diversity data for universal glaucoma visual field grading | |
| Reddy et al. | Evolving a neural network to predict diabetic neuropathy | |
| CN109192312B (en) | Intelligent management system and method for adverse events of heart failure patients | |
| Hsu et al. | Deep learning for automated diabetic retinopathy screening fused with heterogeneous data from EHRs can lead to earlier referral decisions | |
| CN116564521A (en) | Chronic disease risk assessment model establishment method, medium and system | |
| Wang et al. | CNN based multi-instance multi-task learning for syndrome differentiation of diabetic patients | |
| CN114550896A (en) | Dizziness patient emergency pre-examination triage decision-making method, device and model based on artificial neural network | |
| Gong et al. | Practical application of artificial intelligence technology in glaucoma diagnosis |
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 |