CN115116612A - Intelligent risk assessment system and method for child patient state of illness - Google Patents
Intelligent risk assessment system and method for child patient state of illness Download PDFInfo
- Publication number
- CN115116612A CN115116612A CN202210768840.7A CN202210768840A CN115116612A CN 115116612 A CN115116612 A CN 115116612A CN 202210768840 A CN202210768840 A CN 202210768840A CN 115116612 A CN115116612 A CN 115116612A
- Authority
- CN
- China
- Prior art keywords
- child patient
- data
- child
- patient
- knowledge
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
- G06F16/2228—Indexing structures
- G06F16/2255—Hash tables
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/288—Entity relationship models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- 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
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
- G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
- G16H40/67—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for remote operation
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- General Physics & Mathematics (AREA)
- Public Health (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Pathology (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Business, Economics & Management (AREA)
- General Business, Economics & Management (AREA)
- Quality & Reliability (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses an intelligent risk assessment system and method for the state of an illness of a child patient, which comprises the steps of firstly collecting data related to the state of an illness from a family part side and a hospital side of the child patient; then extracting the child patient representation construction entity nodes related to the illness state from the designed data acquisition table, and constructing a child patient illness state risk factor knowledge graph according to the relationship between the expert knowledge and the medical knowledge base construction entities; and then establishing child patient illness state dynamic attribute maps at multiple moments based on the acquired data and the child patient illness state risk factor knowledge maps, learning the attribute maps by using a dynamic heteromorphic graph neural network model, outputting predicted values of important characteristics, evaluating the degree of illness danger based on the predicted values of the important characteristics, and suggesting the treatment strategies of the child patients. The invention uses the historical data and the current data, ensures the accuracy of evaluation of the children patients, focuses on analyzing and predicting the characterization of the children patients, avoids the risk of misdiagnosis and relieves the ethical problem of intelligent medical treatment.
Description
Technical Field
The invention relates to the field of child monitoring, in particular to an intelligent disease risk assessment system and method for a child patient.
Background
Because the pediatric diseases have the characteristics of urgent onset, rapid development, poor regulation capability and the like, the family members usually do not pay attention to the condition of the child patient at the beginning of the condition of the disease, and the hospitalizing process takes long time and is complicated, so that the behavior of the family members carrying the child patient to seek medical treatment when the symptoms of the child patient are not serious is further hindered. On the other hand, some family members are too sensitive to the illness state of children, and occupy medical resources when the children do not need to be hospitalized. Therefore, the preliminary danger early warning and decision-making suggestion before hospitalizing are very important for the children patients, the nonuniformity of medical resource distribution can be relieved, the optimized integration of medical data is realized, and the accuracy, convenience and intellectualization of the medical industry are realized.
The explosive development of big data technology and communication technology provides a new idea for the above problems, and can form a medical service mode taking patient data as a center, and realize an interaction mode between a patient and medical staff, medical institutions and the like, namely intelligent medical treatment. On one hand, under the promotion of the development of big data technology, the mass data generated by the hospital information system realizes the high-efficiency utilization of the data through data processing, data mining and other modes, and the intelligent diagnosis and the medical decision suggestion of the patient are realized by combining medical big data mining, artificial intelligence technology and the like. On the other hand, the characteristics of high rate, low time delay, high reliability and the like of 5G promote the construction of intelligent medical treatment, and can ensure the efficient interaction between the child patient and a medical institution.
However, there are challenges with medical ethics issues facing intelligent medicine. Since the intelligent medical treatment has uncertainty including uncertainty of data analysis and artificial intelligence, the algorithm result is deviated from the real situation, and the physical and monetary losses of the patient are caused. If the wisdom medical treatment misdiagnoses, the attribution of medical responsibility and legal responsibility is disputed and the balance between humanity and science and technology can be damaged. Human life data are very complicated, and professional doctors have misdiagnosis, not to mention artificial intelligence which is not completely industrialized and mature, so that intelligent medical treatment is questioned by many people, and the development of the intelligent medical treatment is seriously hindered.
On the other hand, most applications of smart medicine are directed to the general public, i.e., the population of the whole age group, and research on children is less. However, children do not have independent medical ability and have poor expression ability compared with adults, and family members are easy to neglect the illness condition of children patients and delay medical treatment. Therefore, children need critical early warning more than adults, and intelligent technologies for information acquisition, information fusion, information mining and historical information utilization of children diseases are more needed. Under the condition that children cannot accurately describe own diseases like adults and corresponding answers are carried out on doctor inquiry questions, the degree of danger of children can be evaluated through data mining and analysis, and the attention of family members and doctors is aroused.
Disclosure of Invention
The purpose of the invention is as follows: based on the importance and the research deficiency of pediatric intelligent medical treatment and the ethical problem of intelligent medical treatment, the invention aims to provide the intelligent risk assessment system and the intelligent risk assessment method for the illness state of the child patient, which are used for assessing the illness state danger degree of the child patient and recommending medical decision by collecting, analyzing and mining data of the child patient.
The technical scheme is as follows: in order to achieve the above object, the present invention provides an intelligent risk assessment system for child patient illness state, comprising:
the system comprises a data acquisition module, a child patient medical feature analysis module and a child patient illness state prediction and decision module;
the data acquisition module is used for acquiring data related to illness state from family members and hospitals of the child patients, preprocessing the data, storing the preprocessed data and supporting calling of historical data of the child patients;
the child patient illness state representation data analysis module is used for extracting child patient representation construction entity nodes related to illness states from a designed data acquisition table, and constructing a child patient illness state risk factor knowledge graph according to the relation between expert knowledge and medical knowledge base construction entities; calculating the association degree between the entities to verify and perfect the relationship between the entities;
the child patient illness state prediction and decision-making module is used for constructing child patient illness state dynamic attribute maps at multiple moments based on the acquired data and the child patient illness state risk factor knowledge maps, then utilizing the dynamic abnormal pattern neural network model to learn the attribute maps, outputting predicted values of important characteristics, evaluating the illness degree based on the predicted values of the important characteristics, and suggesting the hospitalization strategies of the child patients; the adjacency matrix of the attribute map of the input dynamic abnormal composition neural network model is determined according to the topological structure of the knowledge map provided by the child patient condition representation data analysis module, each relationship type in the knowledge map corresponds to one adjacency matrix, and the characteristic matrix is determined according to the data acquired by the data acquisition module at each moment; the important characteristics are the characteristics of the disease condition which is screened out and has the influence on the disease condition with the importance degree greater than a set threshold value.
Optionally, the data acquisition module comprises a data acquisition unit of family members of the child patient, a data acquisition unit of a hospital side, a historical data storage and calling unit and a data preprocessing unit;
the family data acquisition unit of the child patient is used for acquiring disease condition related data provided by family, wherein the disease condition related data comprises the demographic characteristics of the child patient, the disease condition, and the physical signs which can be measured by the hospital and the family;
the hospital side data acquisition unit is used for acquiring disease condition related data provided by a hospital, including objective and subjective body characteristics of a child patient, and is divided into in-vitro detection data and body fluid detection data;
the objective in-vitro detection is to detect the body surface of the child patient by using an instrument, the subjective in-vitro detection is based on the evaluation of the physical signs of the child patient by a main doctor, and the objective body fluid detection is based on blood drawing or body fluid of other parts for inspection delivery;
the historical data storage and calling unit is used for establishing an index in a database based on the identity information of the child patient, storing the information of the child patient at a position corresponding to the index according to the acquired time, and forming a data table of a child patient data time sequence with the identity information of the child patient as the index;
and the data preprocessing unit is used for carrying out data cleaning, data conversion and data padding on the collected data.
Optionally, the child patient condition characterization data analysis module comprises a condition risk factor knowledge graph construction unit, a knowledge graph depth analysis unit and a knowledge graph depth fusion unit;
the disease condition risk factor knowledge graph construction unit is used for extracting entity nodes from the family data acquisition table and the hospital data acquisition table of the child patient, establishing an edge relation between the nodes and constructing a knowledge graph; the extraction entity node is a name of body characteristics which influence the illness state of the child patient or possibly bring risks to the health of the child patient, and the characteristics are classified into four types of nodes of life characteristics, acid/alkali characteristics, biochemical detection characteristics and blood characteristics of the cardiovascular/nervous system according to medical classification;
establishing the edge relation of the nodes is based on expert knowledge and a medical knowledge base, modeling the mutual influence of the extracted body characteristics, and connecting the nodes which mutually influence or represent the same disease state;
the knowledge map depth analysis unit is used for calculating the association degree between entity nodes by using an association analysis algorithm based on medical big data of the child patient stored in the database, determining the existence of edges in the knowledge map, and deleting the edges with the association degree lower than a set threshold;
the knowledge map depth fusion unit is used for updating the knowledge map, and comprises data fusion and knowledge fusion; the data fusion is to update the node number and/or type of the child patient illness state risk factor knowledge graph based on the updated family data acquisition table and hospital data acquisition table of the child patient; the knowledge fusion is to update the number and/or types of edges of the child patient condition risk factor knowledge graph based on the updated expert knowledge and the medical knowledge base.
Optionally, the child patient condition prediction and decision module comprises a child patient condition dynamic attribute map construction unit, a child patient condition prediction model unit and a child patient condition evaluation unit;
the child patient illness state dynamic attribute map building unit is used for building a child patient illness state dynamic attribute map, and comprises child patient illness state characteristics at the current moment and child patient illness state characteristics at the past moment; the disease condition representation attribute map of the child patient at the current moment is established according to the disease condition risk factor knowledge map of the child patient, wherein the value of the attribute comprises the specific value or severity description of the corresponding representation; attribute values in the child patient disease condition characterization attribute map at the past moment are provided by child patient historical data, and are stacked and combined with the child patient disease condition characterization attribute map at the current moment by a time sequence to form a dynamic map;
the child patient illness state prediction model unit is used for inputting the child patient illness state dynamic attribute map into the child patient illness state intelligent diagnosis model to predict the value of the important representation of the child patient, wherein the child patient illness state intelligent diagnosis model is a dynamic heterogeneous map neural network model trained by using historical data provided by the data acquisition module;
the important characteristics of the child patient are screened out according to a characteristic selection model based on machine learning, the characteristic selection model is obtained through historical data training of the child patient, the degree of illness danger of the child patient is used as a dependent variable, the characteristics of the child patient are used as independent variables, the relation between the independent variables and the dependent variables is modeled, parameter values are given to the independent variables, and finally the important degree of the independent variables is measured according to the size of the parameter values;
the child patient illness state evaluation unit is used for scoring the illness degree of the child patient based on the predicted value of the important representation of the child patient and in combination with a child critical patient scoring table, and giving immediate hospitalization or selecting a hospitalization suggestion according to the scoring; if the children patient is recommended to seek medical advice immediately, the map interface is connected according to the objective conditions of the children patient to search for the traffic conditions of nearby hospitals and routes and recommend the hospitals.
The objective conditions of the child patients are provided by family data acquisition tables of the child patients, and comprise whether the child patients visit a hospital, the places of the child patients and whether the child patients are transported conditionally;
optionally, the child patient condition evaluation unit is further configured to send the predicted values of the important characteristics of the child patient and the collected data to a doctor in a hospital selected by the family for remote diagnosis, under the condition that the family agrees and does not violate the privacy of the child patient.
The invention provides an intelligent risk assessment method for the condition of a child patient, which comprises the following steps:
collecting data related to illness states from a family part and a hospital part of the child patient, preprocessing and storing the data, wherein the stored data comprises collected historical data of the child patient at different moments;
extracting child patient representation construction entity nodes related to the illness state from a designed data acquisition table, and constructing a child patient illness state risk factor knowledge graph according to the relation between expert knowledge and medical knowledge base construction entities; and calculating the degree of association between the entities to verify and perfect the relationship between the entities;
establishing disease condition dynamic attribute maps of the child patients at multiple moments based on the acquired data and the disease condition risk factor knowledge maps of the child patients, learning the attribute maps by using a dynamic heteromorphic graph neural network model, outputting predicted values of important characteristics, evaluating disease risk degrees based on the predicted values of the important characteristics, and suggesting the treatment strategies of the child patients; wherein, the adjacency matrix of the attribute map of the input dynamic heteromorphic image neural network model is determined according to the topological structure of the knowledge map provided by the child patient disease condition representation data analysis module, each relationship type in the knowledge map corresponds to one adjacency matrix, and the characteristic matrix is determined according to the data at each moment acquired by the data acquisition module; the important characteristics are the characteristics of the disease condition which is screened out and has the influence on the disease condition with the importance degree greater than a set threshold value.
The invention provides a computer system which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the computer program realizes the steps of the intelligent risk assessment method for the condition of the child patient when being loaded to the processor.
Has the advantages that: according to the intelligent risk assessment system and method for the child patient illness state, provided by the invention, on one hand, a novel child patient data analysis method is provided by utilizing data mining and knowledge map technology, the relationship between medical big data and a medical knowledge base can be effectively and dynamically combined, the modeling data is modeled, and new knowledge is mined through correlation analysis. The constructed children patient condition risk factor knowledge graph not only helps medical staff understand and analyze data in a visual mode, but also integrates unstructured data into structured data in a graph mode for an intelligent prediction model. In the intelligent prediction of the illness state of the infant, the method comprehensively uses historical data and current data, avoids the influence of the infant on the fuzzy description of the current illness state of the infant, and ensures the accuracy of evaluation on the infant. On the other hand, the invention also provides a method for avoiding the risk of misdiagnosis and relieving the ethical problem of intelligent medical treatment. Due to the high complexity of human life data and the uncertainty of artificial intelligence, intelligent medical treatment is prone to misdiagnosis and causes social problems. The invention focuses on analyzing and predicting the characteristics of the patient, does not directly make clinical diagnosis with higher difficulty, but makes disease risk early warning for the patient according to the predicted value of the important characteristics and sends the patient data to a doctor for remote diagnosis. Therefore, the risk of misdiagnosis is avoided, the ethical problem of intelligent medical treatment is relieved, and information in artificial intelligence mining data can be used for helping children patients to seek medical treatment in time, relieving medical resource shortage and assisting doctors in diagnosis.
Drawings
In order to more clearly illustrate the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a schematic diagram of an architecture of a system for intelligent risk assessment of a condition of a child patient according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart diagram of a method for acquiring data of a child patient in an embodiment of the invention;
FIG. 3 is a flow chart of a method for analyzing medical characteristics of a pediatric patient according to an embodiment of the invention;
FIG. 4 is a flow chart illustrating a method for predicting and making a decision on a condition of a pediatric patient according to an embodiment of the present invention;
fig. 5 is a flowchart illustrating an intelligent risk assessment method for a condition of a child patient according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, an intelligent risk assessment system for child patient illness disclosed in the embodiments of the present invention mainly includes: the system comprises a data acquisition module, a child patient medical feature analysis module and a child patient illness state prediction and decision module; wherein the content of the first and second substances,
(1) the data acquisition module is used for acquiring data related to illness state from child patient data provided by child patient family and hospitals, preprocessing the data, and reasonably storing and calling historical data of the child patients; the child patient data provided by the child patient family part is based on the demographic characteristics, the disease condition and the simple child patient physical signs which can be measured by the hospitalizing hospital and the family and are acquired by the designed child patient family data acquisition table, and the child patient data provided by the hospital is divided into in-vitro detection and body fluid detection based on the objective and subjective body characterization of the child patient which is acquired by the designed hospital data acquisition table; the storage and calling of the child patient historical data are used for storing the collected child patient data in a background database according to the collection time to form a child patient medical big database, and calling the child patient historical data to assist diagnosis when the child patient diagnoses;
optionally, in an embodiment of the present invention, the child patient data acquisition module at least includes a child patient family data acquisition unit, a hospital data acquisition unit, a data preprocessing unit, and a historical data storage and calling unit;
the data specifically collected by the data collection unit of the family members of the child patients comprise the demographic characteristics of the child patients, the hospital for treatment, the illness condition and the physical signs of the simple child patients. Wherein the demographic characteristics of the child patient comprise age, sex, location and the like; the hospital for seeing a doctor comprises the hospital for seeing a doctor, the grade of the hospital, the area where the hospital is located and the like of the child patient at present; the diseased condition comprises hospitalization diagnosis, child patient hospitalization department, whether the child patient is postoperative, and the like; simple child patient signs include body temperature, heart rate, whether the limb is cold, etc. If the family input of the data of the hospital is that whether the child patient visits the hospital at present, the data of the hospital side needs to be acquired, otherwise, the data is not needed;
the data collected by the hospital side data acquisition unit specifically comprises objective in-vitro detection, subjective in-vitro detection and objective body fluid detection of the child patient, wherein the objective in-vitro detection refers to detection of the body surface of the child patient by using an instrument, such as systolic pressure, diastolic pressure, respiration and the like; subjective in vitro testing is based on the assessment by the attending physician of the patient's physical signs, such as cyanosis, capillary refilling time and the degree of pupillary light reflex; objective body fluid testing is based on blood drawing or other body fluids at other locations for diagnostic testing, such as blood glucose, alkali residual, calcium ion concentration, etc.;
the data preprocessing unit processes incomplete, inconsistent and wrong data in the original data collected by the unit, including data cleaning, conversion and filling, relieves the influence of the abnormalities on a child patient illness state evaluation method, and converts the original data into a form easy for data mining;
the historical data storage and calling unit is used for storing data of all child patients and calling data of all moments of a certain child patient, the storage function codes child patient identity information into Hash codes as indexes, the collected moments are stored in a background SQL database as sequences, and the calling function codes the Hash codes according to the child patient identity information and queries the data of all moments of the child patient in the database by taking the Hash codes as the indexes.
Specifically, the intelligent risk assessment architecture for the condition of the child patient further comprises:
(2) the child patient condition characterization data analysis module is based on a knowledge graph technology, combines the collected child patient condition characterization data and medical expert knowledge analysis data to establish a child patient condition risk factor knowledge graph, and performs depth analysis and depth fusion on the knowledge graph according to the child patient data and a medical knowledge base;
optionally, in an embodiment of the present invention, the child patient condition characterization data analysis module at least includes a condition risk factor knowledge graph construction unit, a knowledge graph depth analysis unit, and a knowledge graph depth fusion unit;
the disease condition risk factor knowledge map construction unit constructs a knowledge map by using a knowledge map construction method based on the data of the child patient and the medical expert knowledge. Wherein the knowledge-graph is composed of triples of factors related to the condition of a patient and influence relationships between the factors based on the knowledge of the medical expert. The factors related to the illness state are based on the characterization names which are extracted from data collected by family members and hospital sides of the child patients and are related to the illness state of the child patients; the influence relationship among the factors refers to the correlation among medical characteristics, wherein a plurality of characteristics have influence on the same aspect of the body of the child patient or one characteristic has influence on a plurality of aspects of the body of the child patient;
the knowledge map depth analysis unit verifies the constructed children patient illness state risk factor knowledge map by using an association algorithm based on the collected children patient medical big database, determines the existence of each node and edge relation in the knowledge map, and perfects the association relation among all entities. The child patient medical big database is the data of all child patients stored in the background by the historical data storage and calling unit; the correlation algorithm is used for calculating Pearson correlation coefficients among nodes based on medical big data of the child patient as the correlation degrees of the nodes; checking whether the edges between the nodes are correct or not based on the association degree between the nodes and a relation threshold value, and deleting if the edges are lower than a set threshold value;
and the knowledge graph depth fusion unit updates the nodes and edges of the knowledge graph, including data fusion and knowledge fusion. The data fusion is based on the updated data acquisition table of the child patient, and the number and/or types of nodes of the child patient condition risk factor knowledge graph are updated; knowledge fusion under the condition that professional knowledge sources are more and more abundant, different source knowledge systems with differences are fused together, and the number and/or types of edges of the disease condition risk factor knowledge graph of the child patient are updated based on updated expert knowledge and a medical knowledge base.
Specifically, the intelligent risk assessment architecture for the condition of the child patient further comprises:
(3) the child patient condition prediction and decision-making module is used for constructing child patient condition dynamic attribute maps at multiple moments based on collected child patient data and child patient condition risk factor knowledge maps, utilizing a child patient condition intelligent diagnosis model to learn the attribute maps and outputting predicted values of important characteristics, evaluating the degree of danger of the child patient based on the predicted values of the important characteristics, recommending the child patient to see a doctor according to results, integrating child patient data and sending the data to a doctor for remote diagnosis.
The child patient illness state dynamic attribute map is formed by combining a child patient illness state representation map at the current moment and a child patient illness state representation map at the past moment. The disease condition representation chart of the child patient at the current moment is established according to the disease condition risk factor knowledge graph of the child patient, wherein the value of the attribute comprises a specific value or severity description of the corresponding representation; the attribute values in the child patient disease condition characterization attribute map at the past moment are provided by the historical data storage and calling unit, the child patient disease condition characterizations at all the moments recorded in the recent period are called to construct an attribute map, and the attribute map and the child patient disease condition characterization attribute map at the current moment are stacked and combined into a dynamic map in a time sequence;
the intelligent child patient condition diagnosis model is a child patient characterization prediction model constructed based on a dynamic heterogeneous graph neural network, takes a child patient condition dynamic attribute graph as input, and predicts a predicted value of future important characterization of a child patient by using current data and past data of the child patient. The important characteristic prediction is important physical signs which are screened out based on a characteristic selection model and have large influence on the disease risk condition of the child patient; the feature selection model utilizes a child patient medical big database stored by a historical data storage and calling unit to train and test the model in the background, and then calls the trained model and parameters to predict the input child patient information;
the evaluation of the disease condition of the child patient comprises evaluation of the disease risk of the child patient and prediction of the disease condition. The evaluation of the danger is based on a predicted value of important characteristics of the child patient, a scoring method of the child critical case is combined to score the child patient and give a suggestion whether the child patient should be admitted immediately, an outpatient registration department suggestion is combined to provide a name and an address of a recommended admission hospital according to objective conditions of the child patient and an external map software interface. The disease condition prediction is that under the condition that the symbolized family members agree and the privacy of the child patient is not involved, the predicted value of the important characteristics of the child patient and the collected data are integrated into an electronic medical record and are sent to a recommended hospital doctor for remote diagnosis, and an interactive communication window between the doctor and the family members of the children is provided.
The following describes a specific process of an intelligent risk assessment method for a child patient's condition according to an embodiment of the present invention with reference to fig. 2 to 5, which mainly includes collecting data related to the child patient's condition, analyzing medical characteristics of the child patient, and predicting and deciding the child patient's condition. As shown in fig. 2, the acquiring of the data related to the condition of the child patient specifically includes:
Specifically, the family members of the child patients fill in a data acquisition table of the family members of the child patients in the form of APP or web pages, and the table acquires comprehensive, reliable and available raw data, including demographic characteristics, hospital visits, illness conditions and simple physical sign data of the child patients.
Wherein the demographic characteristics of the child patient comprise personal information, family condition and location of the child patient, such as name, age, location address and the like, and particularly, the data is subjected to privacy processing, and the privacy data is set to be invisible by using a mask and is only used for model training in the background.
The hospital for seeing a doctor refers to the condition of the current hospital for seeing a doctor of the child patient, and comprises the condition whether the child patient sees the doctor at present, the grade of the hospital, the region where the child patient is located and the like.
The disease condition refers to the condition of the child patient diagnosed at present, including the admission diagnosis of the child patient, the admission of the child patient to a department, whether the child patient is after an operation, etc., and particularly, if the child patient does not visit a hospital at present, the data is not filled in.
The simple child patient physical signs refer to physical sign conditions of the child patient which can be measured under a common family scene, and include body temperature, heart rate, whether limbs are cold or not and the like; if the child patient has a visit in the hospital at present, the data is not filled in, and the physical sign of the child patient provided by the hospital is more accurate than that of the child patient provided by the family.
Specifically, the hospital side at present who has a doctor for the child patient automatically fills in a hospital data acquisition table by introducing pictures of an electronic medical record or a paper medical record through a system by utilizing a regularization matching technology and a character recognition technology, and the table acquires comprehensive, reliable and available original data, including objective in-vitro detection, subjective in-vitro detection and objective body fluid detection of the child patient.
The objective in-vitro detection of the child patient refers to detecting the body surface of the child patient by using an instrument, such as systolic pressure, diastolic pressure, respiration and the like; subjective in vitro testing is based on assessment of the child patient's signs by the attending physician, such as cyanosis, capillary refilling time, and pupil photorefraction; objective body fluid testing is based on blood drawing or other body fluids at other sites for taking tests, such as blood sugar, alkali residual, calcium ion concentration and the like;
and step 130, performing data cleaning, data conversion and data padding on the collected data.
Specifically, because of the default and error conditions of the collected data, the non-uniform units among the data or the unfavorable units for analysis, in order to alleviate the influence of the abnormalities on the disease condition evaluation method of the child patient, the original data is converted into a form easy for data mining.
The data cleaning is to combine the collected data tables and then remove data records with missing data items, for example, when the child patient does not visit a hospital at present, the data of all data items of the hospital side should be removed; when the child patient is in medical treatment at present, the data of the simple physical sign data items are rejected;
the data conversion is to convert each field in the data table according to the field characteristics, recode the redundant discrete data and map the time type data numerically, for example, the admission diagnosis field of the child patient is ' low birth weight and low living ability severe pneumonia ' with pregnancy 28+5 weeks ', and is converted into ' low weight, low living ability and severe pneumonia ';
the data filling is to fill missing data in the record, simple mode filling is adopted for discrete data, mean filling is adopted for numerical data, for example, the characteristic that whether limbs of the child patient get cold is uncertain, the family members can be left unfilled, the data of the child patient is caused to be in default, and in order to solve the problem, the mode of the characteristic of all child patients in the medical big database is called 'no', so that the data of the child patient is filled.
Particularly, the background of the system is provided with a medical big data database for storing data of all children patients, the data acquired by each child patient at each time can be stored after being preprocessed, an acquisition time sequence data table with the information of the child patients as a main index is formed, and the data acquired at all times can be called out according to the information of the child patients when the data of the child patients need to be called. In order to avoid the renaming, the information of the child patient not only comprises the name of the child patient, but also comprises the sex, the birth date and the household location of the child patient, the information is subjected to hash coding, and the obtained code is used as an index. Specifically, the storage step is that the child patient information is subjected to Hash coding, whether the code exists in the main index is searched, and if the code exists, the data acquisition time is used as a secondary index to input the data into a database; if the data does not exist, the code is established as a main code, the data acquisition time is used as an auxiliary index, and the data is recorded into a database. The calling step is that the child patient information is subjected to Hash coding, the position of the code in the main index is searched, and finally the data of the corresponding position is called.
As shown in fig. 3, the medical feature analysis of the pediatric patient specifically includes:
Specifically, all medical data influencing the body of the child patient are divided into four entity types based on collected medical characterization data and a medical knowledge base of the child patient, wherein the four types of nodes comprise life characterization nodes, acid/base characterization nodes, biochemical detection characterization nodes and blood characterization nodes of a cardiovascular/nervous system, the life characterization nodes of the cardiovascular/nervous system comprise systolic blood pressure, heart rate, body temperature, pupillary reflex and the like, the acid/base characterization nodes comprise acidosis, total content of CO2, PH value, PaO2 and the like, the biochemical detection characterization nodes comprise blood sugar, blood potassium, creatinine, blood urea nitrogen and the like, and the blood characterization nodes comprise white blood cell count, prothrombin time, platelet count and the like.
Specifically, based on expert knowledge and medical knowledge base, the above-mentioned partial nodes may affect each other or characterize the same condition, an edge relationship should be established between them, and the edge types should be different between different node types, such as arterial blood carbon dioxide partial pressure in blood characterization not only represents pulmonary ventilation function, but also respiratory factor of body acid-base regulation, if measured value is more than 50mmHg represents respiratory acidosis, and less than 35mmHg represents respiratory acidosis, which has similar effect to the meaning of base surplus in acid/base characterization, they are related and interacting, and should assign a linking edge between the blood characterization and acid/base characterization between the two entity nodes.
And step 230, constructing a child patient illness state risk factor knowledge graph according to the established nodes and edges.
In particular, a knowledge-graph is composed of triples, and one triplet is composed of a head entity (head) and a tail entity (tail), and an association (relationship) between the head entity and the tail entity. And taking the medical condition characterization nodes of the child patients as a head entity and a tail entity, and taking the edges of the relationship between the medical characterizations as an association relationship to form a (head, relation, tail) form triple. And finally, superposing all the triples in the graph to obtain the child patient disease risk factor knowledge graph.
And 240, calculating the association degree among the entity nodes by using an association analysis algorithm based on the medical big data of the child patient stored in the background database, and determining the existence of edges in the knowledge graph.
Specifically, this step aims to verify the original knowledge-graph using the medical big data in order to verify whether the relationship constructed by the medical knowledge base and the expert knowledge is correct. The system firstly calls all child patient data stored in a historical data storage and calling unit, namely child patient medical big data, then calculates the Pearson correlation coefficient between nodes in the big data as the correlation degree between the child patient medical big data, and according to a set threshold value, if the correlation degree between two nodes is lower than the threshold value, the correlation degree between the two nodes is low, the relationship should not exist, and if edges exist between the nodes in the original knowledge graph, the corresponding edges should be deleted.
And step 250, performing data fusion and knowledge fusion on the knowledge graph, and updating nodes and edges in the graph.
Specifically, the purpose of data fusion is to update nodes in the knowledge-graph in the case where the data collection tables have been updated in the data collection module, and the purpose of knowledge fusion is to update edges in the knowledge-graph in the case where the medical knowledge base and expert knowledge are richer.
The data fusion firstly obtains the newly added table entry name of the data collection table, establishes a corresponding node in the graph, searches the type corresponding to the node in the medical knowledge base, searches whether the node is associated with all other nodes or not, if yes, establishes a group of triples, and finally adds the new node and the relationship into the original knowledge graph according to the newly established triples.
The knowledge fusion firstly acquires new knowledge of a certain node, searches whether medical representation related to the new knowledge is used as a node in a knowledge graph or not, if so, detects the relationship between the two nodes, and if no edge exists, establishes an edge; if not, giving up the knowledge and continuing to find the next one.
As shown in fig. 4, the disease prediction and decision of the child patient specifically includes:
and 310, constructing a child patient disease state dynamic attribute map by combining the child patient disease state characteristics at the current moment and the child patient disease state characteristics at the past moment.
Specifically, the child patient condition dynamic attribute map includes all conditions of the child patient within a period of time, not only at the current time, but also data collected at historical times, and is formed by combining the child patient condition characteristics at the current time and the child patient condition characteristics at past times. The attribute graph is input into the intelligent prediction model in the form of a dynamic abnormal graph, and comprises an adjacency matrix and a characteristic matrix of the graph. The adjacent matrix represents a topological structure of the graph, and the knowledge graph is a heterogeneous graph, namely more than one node type and edge type, so that the adjacent matrix comprises R matrixes with dimension of N multiplied by N, N represents the number of nodes in the graph, R represents the number of edge types, if the value of the matrix corresponding to the edge type is 1, the adjacent matrix represents that the edge corresponding to the type exists between two nodes in the graph, and if the value of the edge corresponding to the type is 0, the adjacent matrix represents that the edge corresponding to the type does not exist between two nodes in the graph. The characteristic matrix represents specific values of the nodes, and is obtained by normalizing the specific values of the children patients on medical representation.
And 320, constructing an intelligent disease condition prediction model of the child patient based on the child patient disease condition dynamic attribute graph, and outputting a predicted value of the important characteristics of the child patient.
Specifically, an intelligent child patient condition prediction model is established based on a dynamic heterogeneous graph neural network, the input of the model is a condition dynamic attribute graph of the child patient, the output of the model is a predicted value of important child patient condition characteristics, and parameters of the model are obtained after training is carried out by utilizing large child patient medical data stored in a background.
The dynamic heterogeneous graph neural network is realized by heterogeneous graph neural networks such as RGCN, GAT, and long-short term memory network LSTM. The heterogeneous graph neural network is improved on the traditional graph neural network, the traditional neural network is a homogeneous graph by default, the input only has one adjacency matrix, and the heterogeneous graph has R adjacency matrices, wherein R represents the number of edge types. Therefore, the heteromorphic neural network increases the number of learning parameters, and learns R parameter matrices for R adjacent matrices, respectively. The long-short term memory network is designed for learning the dynamic property of data, and the initial parameter matrix of the heterogeneous graph neural network of the attribute map at a certain moment is calculated by the parameters of the network at the past moment. The specific calculation rule is specified by the LSTM, the parameter matrix at the current moment is calculated from the parameter matrices at all past moments, and the influence of the parameter matrix with shorter time distance on the parameter matrix at the current moment is larger.
The important characteristics of the illness state of the child patient are that characteristics in a dynamic attribute graph of the child patient are screened by using a characteristic selection model in a background, and a part of characteristics which have a large influence on the illness state are selected. The principle of the feature selection algorithm is that input characteristics are screened by using the sparsity of parameters after regular terms are added in machine learning, the condition that the parameter of a certain characteristic is large means that the characteristic is more important to the disease risk degree of a child patient, and otherwise, the condition that the parameter is not important to the disease risk degree of the child patient is shown. The model is trained through medical big data in a background based on a feature selection algorithm, the degree of danger of a child patient is used as a dependent variable, the representation of the child patient is used as an independent variable, parameter values are given to the independent variable, and the relation between the independent variable and the dependent variable is modeled. Finally, the importance degree of the independent variable to the dependent variable can be judged through the parameter value of the independent variable.
And step 330, evaluating the critical degree of the child patient according to the predicted value of the important characteristics of the child patient, deducing the type and the severity of the illness state, and giving medical advice.
Specifically, based on the predicted value of the important characteristics of the illness state of the child patient, the degree of the current physical illness danger of the child patient is scored by combining a plurality of scoring methods for the critical cases of the child patient, and two suggestions of immediate medical treatment and self-selection medical treatment for the child patient can be provided based on the scoring. For example, the disease condition of the child patient can be classified into five grades of serious, general, light and healthy according to the score, if the child patient scores the grade of the general grade and above, the child patient is recommended to see a doctor immediately, and according to the objective condition of the child patient, external map software is called to display the recommended hospital site, the nearby traffic condition and the predicted arrival time, and if the child patient is light or healthy, the child patient is recommended to select to see the doctor by himself. The objective conditions of the child patients are provided by family data acquisition tables of the child patients, and comprise whether the child patients visit a hospital, the places of the child patients and whether the child patients are transported conditionally;
in addition, after the family members decide to go to the hospital, under the condition that the system acquires the consent of the family members, the predicted values of the disease condition characteristics of the current child patients and the important disease condition characteristics of the child patients are made into a structured table form and sent to an outpatient doctor who recommends the hospital, so that remote diagnosis is realized, and the time of the doctor and the family members of the child patients is saved.
The embodiment of the invention discloses a computer system which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the computer program realizes the steps of the intelligent risk assessment method for the condition of the child patient when being loaded to the processor.
It will be understood by those skilled in the art that the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer system (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. The storage medium includes: various media capable of storing computer programs, such as a U disk, a removable hard disk, a read only memory ROM, a random access memory RAM, a magnetic disk, or an optical disk.
Claims (10)
1. An intelligent risk assessment system for the condition of a child patient, comprising:
the data acquisition module is used for acquiring data related to illness state from family members and hospitals of the child patients, preprocessing the data, storing the preprocessed data and supporting calling of historical data of the child patients;
the child patient illness state representation data analysis module is used for extracting child patient representation construction entity nodes related to illness states from a designed data acquisition table, and constructing a child patient illness state risk factor knowledge graph according to the relation between expert knowledge and medical knowledge base construction entities; and calculating the degree of association between the entities to verify and perfect the relationship between the entities;
the child patient illness state prediction and decision-making module is used for constructing child patient illness state dynamic attribute maps at multiple moments based on the acquired data and the child patient illness state risk factor knowledge maps, then utilizing the dynamic abnormal pattern neural network model to learn the attribute maps, outputting predicted values of important characteristics, evaluating the illness danger degree based on the predicted values of the important characteristics, and suggesting the hospitalization strategies of the child patients; wherein, the adjacency matrix of the attribute map of the input dynamic heteromorphic image neural network model is determined according to the topological structure of the knowledge map provided by the child patient disease condition representation data analysis module, each relationship type in the knowledge map corresponds to one adjacency matrix, and the characteristic matrix is determined according to the data at each moment acquired by the data acquisition module; the important characteristics are the characteristics of the disease condition which is screened out and has the influence on the disease condition with the importance degree greater than a set threshold value.
2. The intelligent risk assessment system for pediatric patient condition of claim 1, wherein the data collection module comprises:
the family data acquisition unit of the child patient is used for acquiring disease condition related data provided by family, wherein the disease condition related data comprises the demographic characteristics of the child patient, the disease condition, and the physical signs which can be measured by the hospital and the family;
the hospital side data acquisition unit is used for acquiring disease condition related data provided by a hospital, including objective and subjective body characterization of a child patient, and dividing the data into in-vitro detection data and body fluid detection data;
the historical data storage and calling unit is used for establishing an index in a database based on the identity information of the child patient, storing the information of the child patient at a position corresponding to the index according to the acquired time, and forming a data table of a child patient data time sequence with the identity information of the child patient as the index;
and the data preprocessing unit is used for performing data cleaning, data conversion and data padding on the collected data.
3. The intelligent risk assessment system for condition of a pediatric patient as defined in claim 1, wherein the pediatric patient condition characterization data analysis module comprises:
the disease condition risk factor knowledge graph building unit is used for extracting entity nodes from the family data acquisition table and the hospital data acquisition table of the child patient, building an edge relation between the nodes and building the nodes into a knowledge graph; the extraction entity node is the name of a body representation which influences the illness state of the child patient or possibly brings risks to the health of the child patient, the edge relation of the established node is based on expert knowledge and a medical knowledge base, the mutual influence of the extracted body representations is modeled, and the nodes which mutually influence or represent the same illness state are connected;
the system comprises a knowledge graph depth analysis unit, a data base and a data base, wherein the knowledge graph depth analysis unit is used for calculating the association degree between entity nodes by using an association analysis algorithm based on medical big data of the child patient stored in the data base, determining the existence of edges in a knowledge graph and deleting the edges with the association degree lower than a set threshold value;
the knowledge map depth fusion unit is used for updating the knowledge map, and comprises data fusion and knowledge fusion; the data fusion is based on the updated family data acquisition table and hospital data acquisition table of the child patient, and the number and/or type of nodes of the child patient disease condition risk factor knowledge graph are updated; the knowledge fusion is based on the updated expert knowledge and the medical knowledge base, and the number and/or types of edges of the child patient condition risk factor knowledge graph are updated.
4. The intelligent risk assessment system for pediatric patient condition of claim 1, wherein the pediatric patient condition prediction and decision module comprises:
the child patient illness state dynamic attribute map building unit is used for building a child patient illness state dynamic attribute map, and comprises child patient illness state characteristics at the current moment and child patient illness state characteristics at the past moment; the disease condition representation attribute map of the child patient at the current moment is established according to the disease condition risk factor knowledge map of the child patient, wherein the value of the attribute comprises the specific value or severity description of the corresponding representation; attribute values in the child patient disease condition characterization attribute map at the past moment are provided by child patient historical data, and are stacked and combined with the child patient disease condition characterization attribute map at the current moment by a time sequence to form a dynamic map;
the child patient illness state prediction model unit is used for inputting the child patient illness state dynamic attribute map into a child patient illness state intelligent diagnosis model to predict the value of the important representation of the child patient, wherein the child patient illness state intelligent diagnosis model is a dynamic heterogeneous map neural network model trained by using historical data provided by the data acquisition module;
the child patient illness state evaluation unit is used for scoring the illness degree of the child patient based on the predicted value of the important representation of the child patient and in combination with a child critical patient scoring table, and giving immediate hospitalization or automatically selecting a hospitalization suggestion according to the scoring; if the children patient is recommended to seek medical advice immediately, the map interface is connected according to the objective conditions of the children patient to search for the traffic conditions of nearby hospitals and routes and recommend the hospitals.
5. The intelligent risk assessment system for children's patient condition according to claim 4, wherein the children's patient condition assessment unit is further configured to send the predicted values of the children's patient's vital signs and the collected data to the physicians of the family-selected hospitals for remote diagnosis, with the consent of the family members and without violating the privacy of the children's patients.
6. The intelligent risk assessment system for children patients' conditions according to claim 1, wherein the important characteristics are screened out according to a characteristic selection model based on machine learning, the characteristic selection model is obtained by training historical data of children patients, the degree of risk of children patients is used as a dependent variable, the characteristics of children patients are used as an independent variable, the relationship between the independent variable and the dependent variable is modeled, a parameter value is assigned to the independent variable, and finally the important degree of the independent variable is measured according to the size of the parameter value.
7. An intelligent risk assessment method for the condition of a child patient is characterized by comprising the following steps:
collecting data related to illness states from a family part and a hospital part of the child patient, preprocessing and storing the data, wherein the stored data comprises collected historical data of the child patient at different moments;
extracting child patient representation construction entity nodes related to the illness state from a designed data acquisition table, and constructing a child patient illness state risk factor knowledge graph according to the relation between expert knowledge and medical knowledge base construction entities; and calculating the degree of association between the entities to verify and perfect the relationship between the entities;
establishing disease condition dynamic attribute maps of the child patients at multiple moments based on the acquired data and the disease condition risk factor knowledge maps of the child patients, learning the attribute maps by using a dynamic heteromorphic graph neural network model, outputting predicted values of important characteristics, evaluating disease risk degrees based on the predicted values of the important characteristics, and suggesting the treatment strategies of the child patients; wherein, the adjacency matrix of the attribute map of the input dynamic heteromorphic image neural network model is determined according to the topological structure of the knowledge map provided by the child patient disease condition representation data analysis module, each relationship type in the knowledge map corresponds to one adjacency matrix, and the characteristic matrix is determined according to the data at each moment acquired by the data acquisition module; the important characteristics are the characteristics of the disease condition which is screened out and has the influence on the disease condition with the importance degree greater than a set threshold value.
8. The intelligent risk assessment method for condition of child patient according to claim 7, wherein the construction and updating of the knowledge-graph of condition risk factors of child patient comprises:
extracting entity nodes from a family data acquisition table and a hospital data acquisition table of a child patient, establishing an edge relation between the nodes, and constructing a knowledge graph; the extraction entity node is the name of a body representation which influences the illness state of the child patient or possibly brings risks to the health of the child patient, the edge relation of the established node is based on expert knowledge and a medical knowledge base, the mutual influence of the extracted body representations is modeled, and the nodes which mutually influence or represent the same illness state are connected;
calculating the association degree among entity nodes by using an association analysis algorithm based on medical big data of the child patient stored in a database, determining the existence of edges in a knowledge graph, and deleting the edges with the association degree lower than a set threshold;
updating the node number and/or type of the child patient illness state risk factor knowledge graph based on the updated family data acquisition table and hospital data acquisition table of the child patient; the number and/or type of edges of the child patient condition risk factor knowledge map are updated based on the updated expert knowledge and medical knowledge base.
9. The intelligent risk assessment method for children patients' conditions according to claim 7, wherein the important characteristics predicted by using the dynamic heteromorphic neural network model are screened out according to a characteristic selection model based on machine learning, the characteristic selection model is obtained by training historical data of children patients, the degree of risk of children patients is used as a dependent variable, the characteristics of children patients are used as an independent variable, the relationship between the independent variable and the dependent variable is modeled, a parameter value is given to the independent variable, and finally the important degree of the independent variable is measured according to the size of the parameter value.
10. A computer system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the computer program when loaded into the processor implements the method for intelligent risk assessment of child patient condition according to any of claims 7-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210768840.7A CN115116612A (en) | 2022-07-01 | 2022-07-01 | Intelligent risk assessment system and method for child patient state of illness |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210768840.7A CN115116612A (en) | 2022-07-01 | 2022-07-01 | Intelligent risk assessment system and method for child patient state of illness |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115116612A true CN115116612A (en) | 2022-09-27 |
Family
ID=83331008
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210768840.7A Pending CN115116612A (en) | 2022-07-01 | 2022-07-01 | Intelligent risk assessment system and method for child patient state of illness |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115116612A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116384777A (en) * | 2023-06-06 | 2023-07-04 | 北京科技大学 | Indoor VOCs exposure risk prediction method and device for children group |
| CN117453963A (en) * | 2023-12-26 | 2024-01-26 | 深圳市健怡康医疗器械科技有限公司 | Rehabilitation patient data management system |
| CN118016316A (en) * | 2024-04-10 | 2024-05-10 | 健数(长春)科技有限公司 | Disease screening rate improving method and system by combining knowledge graph with blood routine test data |
-
2022
- 2022-07-01 CN CN202210768840.7A patent/CN115116612A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116384777A (en) * | 2023-06-06 | 2023-07-04 | 北京科技大学 | Indoor VOCs exposure risk prediction method and device for children group |
| CN116384777B (en) * | 2023-06-06 | 2023-08-15 | 北京科技大学 | Indoor VOCs exposure risk prediction method and device for children group |
| CN117453963A (en) * | 2023-12-26 | 2024-01-26 | 深圳市健怡康医疗器械科技有限公司 | Rehabilitation patient data management system |
| CN117453963B (en) * | 2023-12-26 | 2024-03-01 | 深圳市健怡康医疗器械科技有限公司 | Rehabilitation patient data management system |
| CN118016316A (en) * | 2024-04-10 | 2024-05-10 | 健数(长春)科技有限公司 | Disease screening rate improving method and system by combining knowledge graph with blood routine test data |
| CN118016316B (en) * | 2024-04-10 | 2024-06-04 | 健数(长春)科技有限公司 | Disease screening rate improving method and system by combining knowledge graph with blood routine test data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Abdel-Basset et al. | RETRACTED: a novel and powerful framework based on neutrosophic sets to aid patients with cancer | |
| Knaus et al. | The SUPPORT prognostic model: Objective estimates of survival for seriously ill hospitalized adults | |
| CN115116612A (en) | Intelligent risk assessment system and method for child patient state of illness | |
| JP6066826B2 (en) | Analysis system and health business support method | |
| KR20200038628A (en) | Apparatus and method for providing personalized medication information | |
| CN111666477A (en) | Data processing method and device, intelligent equipment and medium | |
| CN114639479A (en) | Intelligent diagnosis auxiliary system based on medical knowledge map | |
| CN112967803A (en) | Early mortality prediction method and system for emergency patients based on integrated model | |
| CN114023441A (en) | Severe AKI early risk assessment model and device based on interpretable machine learning model and development method thereof | |
| CN114783580B (en) | Medical data quality evaluation method and system | |
| CN113035319A (en) | Automatic recommendation method and system for lifestyle management path of diabetic patient | |
| CN112542242A (en) | Data transformation/symptom scoring | |
| KR20110029414A (en) | Method and apparatus for providing health management information | |
| Alghatani et al. | Precision clinical medicine through machine learning: using high and low quantile ranges of vital signs for risk stratification of ICU patients | |
| CN114943629A (en) | Health management and health care service system and health management method thereof | |
| WO2022141925A1 (en) | Intelligent medical service system and method, and storage medium | |
| CN114613487A (en) | Pregnancy decision support system and method | |
| JP7365747B1 (en) | Disease treatment process abnormality identification system based on hierarchical neural network | |
| CN108630290B (en) | Health knowledge personalized recommendation method and system for chronic disease patients | |
| WO2023217737A1 (en) | Health data enrichment for improved medical diagnostics | |
| US20140372146A1 (en) | Determining a physiologic severity of illness score for patients admitted to an acute care facility | |
| Zhao et al. | External validation of a deep learning prediction model for in-hospital mortality among ICU patients | |
| Tolentino et al. | CAREdio: Health screening and heart disease prediction system for rural communities in the Philippines | |
| van der Woerd et al. | Studying sleep: towards the identification of hypnogram features that drive expert interpretation | |
| CN115762698B (en) | Medical chronic disease examination report data extraction method and system |
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 |