CN113871020B - Health management method and system after critical illness diagnosis based on AI machine learning - Google Patents
Health management method and system after critical illness diagnosis based on AI machine learning Download PDFInfo
- Publication number
- CN113871020B CN113871020B CN202111128954.7A CN202111128954A CN113871020B CN 113871020 B CN113871020 B CN 113871020B CN 202111128954 A CN202111128954 A CN 202111128954A CN 113871020 B CN113871020 B CN 113871020B
- Authority
- CN
- China
- Prior art keywords
- health management
- model
- information
- patient
- health
- 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
- G16H50/70—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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- 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
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a health management method and a system after critical illness diagnosis based on AI machine learning, wherein the method comprises the following steps: constructing a health management model of a single disease seed according to the clinical characteristics of the disease seed and the corresponding treatment information, and storing the health management model of the disease seed in a preset model library of an AI machine; confirming whether a target health management model matched with the patient exists according to the case information uploaded by the patient; if yes, performing auxiliary health management on the patient by using the target health management model; and if not, solving the case information to generate a solving result for the AI machine to learn. Therefore, the practicability of the AI machine is improved, the health management model corresponding to the disease category can be effectively provided when various case information is met, and the intelligent and patient experience is improved.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a health management method and system after critical illness treatment based on AI machine learning.
Background
The management after critical treatment is mainly based on manual recording or informatization form recording, a structure consisting of a knowledge system, a language system and a problem processing system is proposed in recent years, the problem processing technology in a relatively mature expert system is introduced into a health management system structure, the situation that the traditional health management is generally lack of knowledge is overcome, a great promotion effect is achieved, the initiative of a user in model construction and model selection cannot be reflected, the closed automatic solving process of the framework system does not have learning capacity, and new knowledge and experience accumulation cannot be learned. This framework system based solely on knowledge is not suitable for complex and realistic follow-up visits and health management. For example, the diagnosis and treatment system based on the expert system can only distinguish and treat diseases within the range owned by the current knowledge base, but if a disease similar to a new crown appears, the diagnosis and treatment system is in no way conditioned, the new crown can not be identified next time, the health management after the intelligent critical illness diagnosis is the product of the combination of follow-up visit after the diagnosis, the health management and Artificial Intelligence (AI), and the research emphasizes on organically combining the knowledge reasoning technology of the AI with the basic functional modules of the health management. In recent years, many hospitals have not been satisfied with simple information processing, and with the development and application of artificial intelligence technology, knowledge has become the most important part of information systems. However, in the use process of the health management system, the user wants to have intelligence, so that the user can accumulate experience in the continuous use process to improve the performance of the user, and therefore, an intelligent after-treatment health management system with a learning function is particularly needed.
Disclosure of Invention
Aiming at the problems shown above, the invention provides a health management method and system after critical illness treatment based on AI machine learning, which are used for solving the problems that the initiative of a user in model construction and model selection cannot be reflected, the closed automatic solving process of the framework system does not have learning capability, and new knowledge and experience accumulation cannot be learned in the background technology.
A health management method after critical illness diagnosis based on AI machine learning comprises the following steps:
constructing a health management model of a single disease seed according to the clinical characteristics of the disease seed and the corresponding treatment information, and storing the health management model of the disease seed in a preset model library of an AI machine;
confirming whether a target health management model matched with the patient exists according to the case information uploaded by the patient;
if yes, performing auxiliary health management on the patient by using the target health management model;
and if not, solving the case information to generate a solving result for the AI machine to learn.
Preferably, the constructing a health management model of a single disease category according to clinical characteristics of the disease category and corresponding treatment information thereof and storing the health management model into a preset model library of the AI machine includes:
generating a post-treatment management plan of a single disease category according to the clinical characteristics of the disease category and the corresponding treatment information;
determining the number of personal follow-up plans and the calling time of each personal follow-up plan according to the post-diagnosis management plan;
constructing an initial model, and training the initial model by using the post-diagnosis management plan, the multiple personal follow-up plans and the calling time of each personal follow-up plan to obtain a health management model of the disease;
and confirming the category of the disease, and storing the health management model into a target partition of a preset model library according to a confirmation result.
Preferably, the confirming whether there is a target health management model matching with the case information uploaded by the patient includes:
receiving and analyzing the case information to generate a problem example;
substituting the problem examples into a preset model library for matching to obtain a plurality of matched first health management models;
analyzing the problem example to obtain hidden characteristic parameters of the problem example, matching the problem example in a plurality of first health management models according to the hidden characteristic parameters, and determining whether a second health management model which is successfully matched exists;
if so, confirming that the target health management model exists, otherwise, confirming that the target health management model does not exist.
Preferably, if present, the performing the assisted health management on the patient using the target health management model comprises:
acquiring an electronic medical record and electronic identity information of a patient;
associating the electronic medical record and the electronic identity information with the target health management model;
acquiring the daily behavior information of a patient, judging whether the calling time of each personal follow-up plan conflicts with the work and rest plan of the patient according to the daily behavior information, if so, automatically adjusting the calling time of each personal follow-up plan, otherwise, not needing to perform subsequent operation;
and controlling the AI machine to carry out personal follow-up and auxiliary health management on the patient according to the target health management model.
Preferably, if the case information does not exist, solving the case information to generate a solution result for the AI machine to learn, including:
confirming the disease type of the patient according to the medical record information;
calling expected health management models of the same type from a preset model library according to the disease types;
solving the case information by using the expected health management model to obtain a first solving result;
and feeding back the first solving result to a patient terminal for evaluation of the patient, confirming whether the first solving result is available information according to the evaluation result, and if so, generating a model control knowledge according to the first solving result and storing the model control knowledge in a preset knowledge base of the AI machine for learning of the AI machine.
Preferably, the method further comprises:
if the first solving result is not available information, searching relevant experience from a preset experience library of the AI machine;
solving the case information by using the related experience to obtain a second solving result;
constructing a brand new health management model according to the second solving result and the case information;
storing the brand new health management model into a preset model library, and generating learning experience according to the second solving result;
and storing the learning experience into the preset experience library.
Preferably, the method further comprises:
detecting the learning progress of the AI machine on model control knowledge and learning experience in real time, and generating test case information when the learning progress of the AI machine on the model control knowledge and the learning experience is finished;
performing model matching test on the AI machine through the test case information to obtain a test result;
and confirming whether the learning effect of the AI machine is qualified or not according to the test result, if so, not needing to carry out subsequent operation, otherwise, inputting the model manipulation knowledge into the expected management model for updating, retraining the updated expected management model and training the brand-new health management model.
Preferably, after the target health management model is utilized to perform the auxiliary health management on the patient, the method further comprises:
acquiring follow-up visit information of a personal follow-up visit plan of a patient, and generating health information representation of the patient according to the follow-up visit information;
acquiring the characteristic representation of each information source in the health information representation, and analyzing the health state of the patient according to the characteristic representation of each information source to acquire an analysis result;
when the analysis result is healthy, subsequent operation is not needed, and when the analysis result is unhealthy, auxiliary decision information of each information source is generated according to the feature representation of the information source;
performing associated modeling on the auxiliary decision information and the treatment information of the patient by utilizing a preset time-space convolutional network;
acquiring first sign factor correlation information of a diseased organ of a patient in a healthy state and second sign factor correlation information of the diseased organ in a current state;
taking the second sign factor correlation information as the input of a correlation model and the first sign factor correlation information as the output of the correlation model, and obtaining a health decision variable output by the correlation model;
constructing a linear relation between the in vitro influence factors of the patient and the health index;
determining the influence degree of the in vitro influence factors corresponding to the health recovery of the patient according to the maximum linear value in the linear relation;
determining whether the influence degree is greater than or equal to a preset threshold value, if so, counting influence factors corresponding to all health decision variables and generating a first follow-up health suggestion of the patient;
otherwise, determining the target association degree between each health decision variable and the in-vitro influence factors, and marking the target health decision variables with the target association degree being more than or equal to the preset association degree;
generating a second subsequent health recommendation of the patient according to the target influence factor corresponding to the target health decision variable;
uploading the first subsequent health advice or the second subsequent health advice to an attending physician terminal for decision making to confirm whether to execute or not, and if so, importing the first subsequent health advice or the second subsequent health advice into the target health management model.
Preferably, the method further comprises:
determining a learning index of the model manipulation knowledge, importing the expected health management model and the learning index into a preset association database for association degree judgment, and acquiring a judgment result;
confirming whether the expected health management model completely learns the model manipulation knowledge or not according to a comparison result of the current relevance and a preset relevance in the judgment result, if so, not needing subsequent operation, and otherwise, extracting data items of each learning index;
setting a specific mark for each data item of the learning index;
acquiring the matching attribute of each data item, matching the matching attribute with the expected health management model, and classifying the target learning indexes which are not successfully matched in the data items into a preset data set according to the matching result;
obtaining the dimension attribute of each target learning index in the preset data set, and generating a training set according to the dimension attribute;
training the desired health management model with the training set until it converges.
An AI machine learning-based post critical care health management system, the system comprising:
the construction module is used for constructing a health management model of a single disease type according to the clinical characteristics of the disease type and the corresponding treatment information and storing the health management model of the disease type in a preset model library of the AI machine;
the confirmation module is used for confirming whether a target health management model matched with the patient exists according to the case information uploaded by the patient;
the auxiliary module is used for performing auxiliary health management on the patient by utilizing the target health management model if the target health management model exists;
and the solving module is used for solving the case information to generate a solving result for the AI machine to learn if the case information does not exist.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
FIG. 1 is a flowchart illustrating a health management method after critical care based on AI machine learning according to the present invention;
FIG. 2 is another flowchart of the health management method after critical care based on AI machine learning according to the present invention;
FIG. 3 is a flowchart illustrating a method for health management after critical care based on AI machine learning according to the present invention;
FIG. 4 is a screenshot of an embodiment of a health management method after critical care based on AI machine learning according to the present invention;
FIG. 5 is a sectional view of another embodiment of a health management method after critical care based on AI machine learning according to the present invention;
fig. 6 is a schematic structural diagram of a health management system after critical care treatment based on AI machine learning according to the present invention.
Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
The management after critical treatment is mainly based on manual recording or informatization form recording, a structure consisting of a knowledge system, a language system and a problem processing system is proposed in recent years, the problem processing technology in a relatively mature expert system is introduced into a health management system structure, the situation that the traditional health management is generally lack of knowledge is overcome, a great promotion effect is achieved, the initiative of a user in model construction and model selection cannot be reflected, the closed automatic solving process of the framework system does not have learning capacity, and new knowledge and experience accumulation cannot be learned. This framework system based solely on knowledge is not suitable for complex and realistic follow-up visits and health management. For example, the diagnosis and treatment system based on the expert system can only distinguish and treat diseases within the range owned by the current knowledge base, but if a disease similar to a new crown appears, the diagnosis and treatment system is in no way conditioned, the new crown can not be identified next time, the health management after the intelligent critical illness diagnosis is the product of the combination of follow-up visit after the diagnosis, the health management and Artificial Intelligence (AI), and the research emphasizes on organically combining the knowledge reasoning technology of the AI with the basic functional modules of the health management. In recent years, many hospitals have not been satisfied with simple information processing, and with the development and application of artificial intelligence technology, knowledge has become the most important part of information systems. However, in the use process of the health management system, the user wants to have intelligence, so that the user can accumulate experience in the continuous use process to improve the performance of the user, and therefore, an intelligent after-treatment health management system with a learning function is particularly needed. In order to solve the above problems, the present embodiment discloses a health management method after critical care based on AI machine learning.
A health management method after critical care treatment based on AI machine learning, as shown in fig. 1, comprising the following steps:
s101, constructing a health management model of a single disease type according to clinical characteristics of the disease type and corresponding treatment information, and storing the health management model of the disease type in a preset model library of an AI machine;
step S102, confirming whether a target health management model matched with the case information uploaded by the patient exists or not according to the case information uploaded by the patient;
step S103, if the target health management model exists, performing auxiliary health management on the patient by using the target health management model;
and step S104, if the case information does not exist, solving the case information to generate a solving result for the AI machine to learn.
The working principle of the technical scheme is as follows: constructing a health management model of a single disease category according to the clinical characteristics of the disease category and the corresponding treatment information, storing the health management model into a preset model library of an AI machine, at the moment, the AI machine can search out the matched target disease species from the preset model library according to the disease characteristics fed back by the patient, then the patient is subjected to auxiliary after-treatment health management according to a target health management model corresponding to the target disease species, the steps are established on the premise that the health management model with known disease species exists in the model base, when an unknown disease species exists, i.e. it is confirmed from the case information uploaded by the patient that there is no target health management model matching it, the case information is solved by a preset method to obtain experience knowledge according to the solving result, and the experience knowledge is used for the AI machine to learn so as to construct a health management model of unknown disease species, so that the AI machine can effectively take measures when meeting the same case information next time.
The beneficial effects of the above technical scheme are: the health management model of each single disease is constructed in advance and stored in the AI machine, so that matching is carried out according to case information provided by a patient to determine whether the matched health management model exists or not, and then the case information is solved after the health management model is not matched to generate empirical knowledge for the learning of the AI machine, so that the AI machine can realize intelligent learning, the practicability is improved, the health management model corresponding to the disease can be effectively provided when various case information is met, the intelligence and the experience of the patient are improved, the problems that the model building of a user cannot be reflected in the prior art, the initiative of the model selection cannot be realized in the prior art, the closed automatic solving process of the framework system also has no learning capacity, and the problems of new knowledge and experience accumulation cannot be learned are solved.
In an embodiment, as shown in fig. 2, the constructing a health management model of a single disease category according to clinical characteristics of the disease category and corresponding treatment information and storing the health management model into a preset model library of an AI machine includes:
step S201, generating a post-treatment management plan of a single disease type according to the clinical characteristics of the disease type and the corresponding treatment information;
step S202, determining the number of the personal follow-up plans and the calling time of each personal follow-up plan according to the post-diagnosis management plan;
step S203, constructing an initial model, and training the initial model by using the post-diagnosis management plan, the multiple personal follow-up plans and the calling time of each personal follow-up plan to obtain a health management model of the disease;
and S204, confirming the category of the disease, and storing the health management model into a target partition of a preset model library according to a confirmation result.
The beneficial effects of the above technical scheme are: the most detailed and scientific health management model of each disease can be obtained by generating a detailed after-treatment management plan of each disease to train an initial model, so that the AI machine can more humanize to carry out auxiliary after-treatment health management on patients according to the health management model of each disease, the intelligence is improved, furthermore, the types of the diseases are distinguished, the models of the same type of diseases can be intensively managed, the applicable models can be quickly matched in the subsequent model calling process, and the working efficiency is improved.
In one embodiment, as shown in fig. 3, the confirming whether there is a target health management model matching with the case information uploaded by the patient includes:
s301, receiving and analyzing the case information to generate a problem example;
step S302, substituting the problem cases into a preset model library for matching to obtain a plurality of matched first health management models;
step S303, analyzing the problem instance to obtain hidden characteristic parameters of the problem instance, matching the problem instance in a plurality of first health management models according to the hidden characteristic parameters, and determining whether a second health management model which is successfully matched exists;
and S304, if yes, confirming that the target health management model exists, and otherwise, confirming that the target health management model does not exist.
The beneficial effects of the above technical scheme are: whether the matched health management model exists in the preset model library or not can be accurately determined by performing secondary matching on the problem example, so that the matching success and accuracy are improved.
In one embodiment, the performing assisted health management of the patient using the target health management model, if any, comprises:
acquiring an electronic medical record and electronic identity information of a patient;
associating the electronic medical record and the electronic identity information with the target health management model;
acquiring the daily behavior information of a patient, judging whether the calling time of each personal follow-up plan conflicts with the work and rest plan of the patient according to the daily behavior information, if so, automatically adjusting the calling time of each personal follow-up plan, otherwise, not needing to perform subsequent operation;
and controlling the AI machine to carry out personal follow-up and auxiliary health management on the patient according to the target health management model.
The beneficial effects of the above technical scheme are: the calling time of the personal follow-up plan is intelligently adjusted, so that the adjusted calling time not only meets the needs of the patient, but also does not influence the daily work and rest plan of the patient, the experience of the patient is further improved, and meanwhile, intelligent auxiliary health management is realized.
In one embodiment, if not, solving the case information to generate a solution result for the AI machine to learn includes:
confirming the disease type of the patient according to the medical record information;
calling expected health management models of the same type from a preset model library according to the disease types;
solving the case information by using the expected health management model to obtain a first solving result;
and feeding back the first solving result to a patient terminal for evaluation of the patient, confirming whether the first solving result is available information according to the evaluation result, and if so, generating a model control knowledge according to the first solving result and storing the model control knowledge in a preset knowledge base of the AI machine for learning of the AI machine.
The beneficial effects of the above technical scheme are: the case information is solved by using the similar expected health management model, the health management model of the similar diseases can be used for solving according to the similar characteristics corresponding to the disease types, the final solving result can be more practical, further, the first solving result is used for the patient to evaluate, whether the first solving result is qualified or not can be effectively determined by combining the actual evaluation content of the patient, whether useful empirical knowledge exists in the first solving result or not can be determined, the effective screening of the empirical knowledge can be realized, and the learning efficiency of the AI machine is improved.
In one embodiment, the method further comprises:
if the first solving result is not available information, searching relevant experience from a preset experience library of the AI machine;
solving the case information by using the related experience to obtain a second solving result;
constructing a brand new health management model according to the second solving result and the case information;
storing the brand new health management model into a preset model library, and generating learning experience according to the second solving result;
and storing the learning experience into the preset experience library.
The beneficial effects of the above technical scheme are: the case information is solved by using the related experience, so that a solution method related to the case information can be obtained to the maximum extent, a brand new health management model is built to deal with the case information, the AI machine can quickly call the brand new health management model to perform auxiliary health management in the subsequent treatment of the case information, the working efficiency is further improved, the AI machine can learn and absorb the learning experience to the maximum extent by storing the learning experience in the preset experience library, and the guarantee is provided for the subsequent matching of the model.
In one embodiment, the method further comprises:
detecting the learning progress of the AI machine on model control knowledge and learning experience in real time, and generating test case information when the learning progress of the AI machine on the model control knowledge and the learning experience is finished;
performing model matching test on the AI machine through the test case information to obtain a test result;
and confirming whether the learning effect of the AI machine is qualified or not according to the test result, if so, not needing to carry out subsequent operation, otherwise, inputting the model manipulation knowledge into the expected management model for updating, retraining the updated expected management model and training the brand-new health management model.
The beneficial effects of the above technical scheme are: the learning effect of the AI machine is tested, so that the calling accuracy and proficiency of the AI machine on the newly added health management model and the updated existing health management model can be effectively determined, and the working efficiency is further improved.
In one embodiment, after performing assisted health management on a patient using the target health management model, the method further comprises:
acquiring follow-up visit information of a personal follow-up visit plan of a patient, and generating health information representation of the patient according to the follow-up visit information;
acquiring the characteristic representation of each information source in the health information representation, and analyzing the health state of the patient according to the characteristic representation of each information source to acquire an analysis result;
when the analysis result is healthy, subsequent operation is not needed, and when the analysis result is unhealthy, auxiliary decision information of each information source is generated according to the feature representation of the information source;
performing associated modeling on the auxiliary decision information and the treatment information of the patient by utilizing a preset time-space convolutional network;
acquiring first sign factor correlation information of a diseased organ of a patient in a healthy state and second sign factor correlation information of the diseased organ in a current state;
taking the second sign factor correlation information as the input of a correlation model and the first sign factor correlation information as the output of the correlation model, and obtaining a health decision variable output by the correlation model;
constructing a linear relation between the in vitro influence factors of the patient and the health index;
determining the influence degree of the in vitro influence factors corresponding to the health recovery of the patient according to the maximum linear value in the linear relation;
determining whether the influence degree is greater than or equal to a preset threshold value, if so, counting influence factors corresponding to all health decision variables and generating a first follow-up health suggestion of the patient;
otherwise, determining the target association degree between each health decision variable and the in-vitro influence factors, and marking the target health decision variables with the target association degree being more than or equal to the preset association degree;
generating a second subsequent health recommendation of the patient according to the target influence factor corresponding to the target health decision variable;
uploading the first subsequent health advice or the second subsequent health advice to an attending physician terminal for decision making to confirm whether to execute or not, and if so, importing the first subsequent health advice or the second subsequent health advice into the target health management model.
The beneficial effects of the above technical scheme are: the recovery condition and the influence factors influencing recovery can be intelligently determined according to the follow-up plan of the patient, and then the follow-up health advice can be intelligently generated to be used for the patient to carry out more scientific and healthy after-treatment health management, so that the experience of the patient is improved, meanwhile, the follow-up health advice suitable for the patient can be accurately generated according to the body parameters and the in-vitro influence parameters of the patient by generating different follow-up health advice, and the practicability is improved.
Preferably, the method further comprises:
determining a learning index of the model manipulation knowledge, importing the expected health management model and the learning index into a preset association database for association degree judgment, and acquiring a judgment result;
confirming whether the expected health management model completely learns the model manipulation knowledge or not according to a comparison result of the current relevance and a preset relevance in the judgment result, if so, not needing subsequent operation, and otherwise, extracting data items of each learning index;
setting a specific mark for each data item of the learning index;
acquiring the matching attribute of each data item, matching the matching attribute with the expected health management model, and classifying the target learning indexes which are not successfully matched in the data items into a preset data set according to the matching result;
obtaining the dimension attribute of each target learning index in the preset data set, and generating a training set according to the dimension attribute;
training the desired health management model with the training set until it converges.
In one embodiment, the method comprises the following steps: the learning effect of the model is accurately evaluated by determining the learning condition of the expected health management model to the model manipulation knowledge, so that the content which is not learned by the model is selected for special learning training, the learning progress of the model is ensured, the learned learning indexes are removed to optimize the memory in the preset knowledge base, the system load is reduced, and the stability and the working efficiency are improved.
Acquiring a patient list, wherein each item in the patient list carries calling time, a user identity identifier, a user communication number, a user personal follow-up plan, a post-diagnosis management plan and an electronic medical record of a user; calling users one by one according to the patient list; and after the user answers the phone, the phone communication is carried out with the user according to the user personal follow-up plan and the electronic medical record, and a doctor is assisted to finish the personalized medical follow-up and after-treatment health management of the user. The invention can relieve the situations of insufficient medical resource supply, less strict after-treatment management and self improvement.
The system comprises a patient terminal, a doctor terminal and a post-treatment management platform, wherein the patient terminal is used for displaying post-treatment management patient end software, the doctor terminal is used for displaying post-treatment management doctor end software, and the post-treatment management platform comprises: a patient information management module; a doctor information management module; a clinical event management module; and the flow driving module is used for receiving the clinical events uploaded by the patient terminal and sending the clinical events to the corresponding doctor users, and receiving the illness state processing information sent by the doctor terminal and sending the illness state processing information to the corresponding patient users.
As shown in fig. 4, the health management architecture after critical care treatment based on AI machine learning includes five components, namely an interactive system, a problem solving system, a data system, a model system and a knowledge system.
1. The interactive system provides a user with an interactive window of the system, provides various information for the user, and provides support for solving a decision problem, extracting and organizing data, building a model, processing knowledge and the like.
2. And the problem solving system is characterized in that the case receiver is responsible for generating a problem case, and a problem solver is used for manipulating knowledge through a model to generate a solution of the problem through deductive reasoning. The analysis evaluation module evaluates the solution according to the condition of the solution and the satisfaction degree of the user, and divides the solution into a positive example or a negative example. And the learning module performs inductive learning on the problem solving process by combining the domain level knowledge according to the analysis result to generate new model control knowledge or correct the original model.
3. The data system comprises a database and a database management system, wherein the database comprises an internal database and an external database, and is uniformly managed by the database management system.
4. The model system comprises a method library, a model dictionary and a corresponding model management system.
5. The knowledge system comprises a multi-attribute knowledge base (facts and rules), a knowledge base management system and an inference component.
As shown in fig. 5, the learning system is mainly composed of four components:
(1) the learning effect is closely related to the environment and knowledge level of the system. The environment may be the subject or an objective condition to which the subject is subjected, such as a decision problem in post-diagnosis health management, a data system, a model system, and the like. No matter what learning method is adopted by the learning system, the quality of the information provided by the environment has a direct influence on the completion degree of the learning task.
(2) The quality of a learning system is closely related to the design of the knowledge system. Knowledge representation and reasoning methods are important points to be considered when designing a learning system. Knowledge in the learning system knowledge base is divided into Long Term Memory (LTM) and Medium Term Memory (MTM). The knowledge content memorized for a long time is the necessary background knowledge of the learning system, and the basic knowledge and the category are stable and basically unchanged in the learning process. The knowledge content memorized in the middle stage is relatively stable, such as the law of various specific things, and the knowledge can be changed through learning.
(3) And a decision-making step which makes a decision or takes an action to complete various works by utilizing knowledge in the knowledge base. The feedback information provided by this link to the learning link is very important, because the learning system needs some evaluation method to evaluate the learning effect and improve the performance of the system accordingly.
(4) The learning link is the core of the whole learning system, and the learning link searches, controls and operates (generalizes, abstracts, reasoning and the like) the environmental information to generate, correct and supplement knowledge.
The beneficial effects of the above technical scheme are: the system is convenient for uploading clinical events of patients and carrying out the work of doctor diagnosis, follow-up visit, information management after the diagnosis and medical assistance, and has wider application range. The method can organically combine the solving process and the learning process of the decision problem, and can continuously improve the performance of the health management system through experience accumulation, so that the health management system after critical illness diagnosis has stronger self-learning capability and adaptability.
This embodiment also discloses health management system after critical illness diagnosis based on AI machine learning, as shown in fig. 6, this system includes:
the building module 601 is used for building a health management model of a single disease category according to the clinical characteristics of the disease category and the corresponding treatment information and storing the health management model of the disease category in a preset model library of the AI machine;
a confirmation module 602, configured to confirm whether a target health management model matching the patient exists according to the case information uploaded by the patient;
an auxiliary module 603, configured to perform auxiliary health management on the patient using the target health management model if the target health management model exists;
and the solving module 604 is configured to solve the case information to generate a solving result for the AI machine to learn if the case information does not exist.
The working principle and the advantageous effects of the above technical solution have been explained in the method claims, and are not described herein again.
It will be understood by those skilled in the art that the first and second terms of the present invention refer to different stages of application.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.
Claims (9)
1. A health management method after critical illness diagnosis based on AI machine learning is characterized by comprising the following steps:
constructing a health management model of a single disease seed according to the clinical characteristics of the disease seed and the corresponding treatment information, and storing the health management model of the disease seed in a preset model library of an AI machine;
confirming whether a target health management model matched with the patient exists according to the case information uploaded by the patient;
if yes, performing auxiliary health management on the patient by using the target health management model;
if the case information does not exist, solving the case information to generate a solving result for the AI machine to learn;
the method for constructing the health management model of the disease category according to the clinical characteristics of the single disease category and the corresponding treatment information and storing the health management model of the disease category in a preset model library of an AI machine comprises the following steps:
generating a post-treatment management plan of a single disease category according to the clinical characteristics of the disease category and the corresponding treatment information;
determining the number of personal follow-up plans and the calling time of each personal follow-up plan according to the post-diagnosis management plan;
constructing an initial model, and training the initial model by using the number of the post-diagnosis management plans and the number of the individual follow-up plans and the calling time of each individual follow-up plan to obtain a health management model of the disease;
and confirming the category of the disease, and storing the health management model into a target partition of a preset model library according to a confirmation result.
2. The AI machine learning-based post-critical care health management method according to claim 1, wherein the determining from the case information uploaded by the patient whether there is a matching target health management model comprises:
receiving and analyzing the case information to generate a problem example;
substituting the problem examples into a preset model library for matching to obtain a plurality of matched first health management models;
analyzing the problem example to obtain hidden characteristic parameters of the problem example, matching the problem example in a plurality of first health management models according to the hidden characteristic parameters, and determining whether a second health management model which is successfully matched exists;
if so, confirming that the target health management model exists, otherwise, confirming that the target health management model does not exist.
3. The AI machine learning-based post-critical care health management method of claim 1, wherein the performing assisted health management of the patient using the target health management model, if any, comprises:
acquiring an electronic medical record and electronic identity information of a patient;
associating the electronic medical record and the electronic identity information with the target health management model;
acquiring the daily behavior information of a patient, judging whether the calling time of each personal follow-up plan conflicts with the work and rest plan of the patient according to the daily behavior information, if so, automatically adjusting the calling time of each personal follow-up plan, otherwise, not needing to perform subsequent operation;
and controlling the AI machine to carry out personal follow-up and auxiliary health management on the patient according to the target health management model.
4. The AI machine learning-based health management method after critical care, according to claim 1, wherein if not, solving the case information to generate a solution result for the AI machine to learn, comprising:
confirming the disease type of the patient according to the case information;
calling expected health management models of the same type from a preset model library according to the disease types;
solving the case information by using the expected health management model to obtain a first solving result;
and feeding back the first solving result to a patient terminal for evaluation of the patient, confirming whether the first solving result is available information according to the evaluation result, and if so, generating a model control knowledge according to the first solving result and storing the model control knowledge in a preset knowledge base of the AI machine for learning of the AI machine.
5. The AI machine learning-based post critical care health management method of claim 4, further comprising:
if the first solving result is not available information, searching relevant experience from a preset experience library of the AI machine;
solving the case information by using the related experience to obtain a second solving result;
constructing a brand new health management model according to the second solving result and the case information;
storing the brand new health management model into a preset model library, and generating learning experience according to the second solving result;
and storing the learning experience into the preset experience library.
6. The AI machine learning-based post critical care health management method of claim 4 or 5, further comprising:
detecting the learning progress of the AI machine on model control knowledge and learning experience in real time, and generating test case information when the learning progress of the AI machine on the model control knowledge and the learning experience is finished;
performing model matching test on the AI machine through the test case information to obtain a test result;
and confirming whether the learning effect of the AI machine is qualified or not according to the test result, if so, not needing to carry out subsequent operation, otherwise, inputting the model manipulation knowledge into the expected health management model for updating, retraining the updated expected health management model and training the brand-new health management model.
7. The AI machine learning-based post critical care health management method of claim 1, wherein after performing assisted health management of a patient using the target health management model, the method further comprises:
acquiring follow-up visit information of a personal follow-up visit plan of a patient, and generating health information representation of the patient according to the follow-up visit information;
acquiring the characteristic representation of each information source in the health information representation, and analyzing the health state of the patient according to the characteristic representation of each information source to acquire an analysis result;
when the analysis result is healthy, subsequent operation is not needed, and when the analysis result is unhealthy, auxiliary decision information of each information source is generated according to the feature representation of the information source;
performing associated modeling on the auxiliary decision information and the treatment information of the patient by utilizing a preset time-space convolutional network;
acquiring first sign factor correlation information of a diseased organ of a patient in a healthy state and second sign factor correlation information of the diseased organ in a current state;
taking the second sign factor correlation information as the input of a correlation model and the first sign factor correlation information as the output of the correlation model, and obtaining a health decision variable output by the correlation model;
constructing a linear relation between the in vitro influence factors of the patient and the health index;
determining the influence degree of the in vitro influence factors corresponding to the health recovery of the patient according to the maximum linear value in the linear relation;
determining whether the influence degree is greater than or equal to a preset threshold value, if so, counting influence factors corresponding to all health decision variables and generating a first follow-up health suggestion of the patient;
otherwise, determining the target association degree between each health decision variable and the in-vitro influence factors, and marking the target health decision variables with the target association degree being more than or equal to the preset association degree;
generating a second subsequent health recommendation of the patient according to the target influence factor corresponding to the target health decision variable;
uploading the first subsequent health advice or the second subsequent health advice to an attending physician terminal for decision making to confirm whether to execute or not, and if so, importing the first subsequent health advice or the second subsequent health advice into the target health management model.
8. The AI machine learning-based post critical care health management method of claim 4, further comprising:
determining a learning index of the model manipulation knowledge, importing the expected health management model and the learning index into a preset association database for association degree judgment, and obtaining a judgment result;
confirming whether the expected health management model completely learns the model manipulation knowledge or not according to a comparison result of the current relevance and a preset relevance in the judgment result, if so, not needing subsequent operation, and otherwise, extracting data items of each learning index;
setting a specific mark for each data item of the learning index;
acquiring the matching attribute of each data item, matching the matching attribute with the expected health management model, and classifying the target learning indexes which are not successfully matched in the data items into a preset data set according to the matching result;
obtaining the dimension attribute of each target learning index in the preset data set, and generating a training set according to the dimension attribute;
training the desired health management model with the training set until it converges.
9. A health management system after critical care treatment based on AI machine learning, the system comprising:
the construction module is used for constructing a health management model of a single disease type according to the clinical characteristics of the disease type and the corresponding treatment information and storing the health management model of the disease type in a preset model library of the AI machine;
the confirmation module is used for confirming whether a target health management model matched with the patient exists according to the case information uploaded by the patient;
the auxiliary module is used for performing auxiliary health management on the patient by utilizing the target health management model if the target health management model exists;
the solving module is used for solving the case information to generate a solving result for the AI machine to learn if the case information does not exist;
the step that the construction module constructs the health management model of the disease species according to the clinical characteristics of the single disease species and the corresponding treatment information thereof and stores the health management model into the preset model library of the AI machine comprises the following steps:
generating a post-treatment management plan of a single disease category according to the clinical characteristics of the disease category and the corresponding treatment information;
determining the number of personal follow-up plans and the calling time of each personal follow-up plan according to the post-diagnosis management plan;
constructing an initial model, and training the initial model by using the post-diagnosis management plan, the multiple personal follow-up plans and the calling time of each personal follow-up plan to obtain a health management model of the disease;
and confirming the category of the disease, and storing the health management model into a target partition of a preset model library according to a confirmation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111128954.7A CN113871020B (en) | 2021-09-26 | 2021-09-26 | Health management method and system after critical illness diagnosis based on AI machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111128954.7A CN113871020B (en) | 2021-09-26 | 2021-09-26 | Health management method and system after critical illness diagnosis based on AI machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113871020A CN113871020A (en) | 2021-12-31 |
| CN113871020B true CN113871020B (en) | 2022-04-01 |
Family
ID=78994491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111128954.7A Active CN113871020B (en) | 2021-09-26 | 2021-09-26 | Health management method and system after critical illness diagnosis based on AI machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113871020B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115153468B (en) * | 2022-07-11 | 2023-04-07 | 广州医科大学附属第一医院（广州呼吸中心） | Chronic disease follow-up visit monitoring system based on two-dimensional code |
| CN115579132B (en) * | 2022-12-09 | 2023-08-22 | 智慧中医科技(广东)有限公司 | Verification method, device and system based on traditional Chinese medicine diagnosis and treatment experience file format |
| CN116936078B (en) * | 2023-07-18 | 2024-04-26 | 南京大经中医药信息技术有限公司 | Traditional Chinese medicine pre-inquiry collection management system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU5355199A (en) * | 1996-09-30 | 1999-12-02 | Smithkline Beecham Corporation | Disease management method and system |
| CN109147943A (en) * | 2018-06-28 | 2019-01-04 | 刘毅 | System is assessed and instructed to perioperative |
| CN111383770A (en) * | 2020-03-09 | 2020-07-07 | 北京大学 | Clinical research and follow-up visit fusion system and method |
| CN113506630A (en) * | 2021-07-08 | 2021-10-15 | 上海中医药大学附属龙华医院 | Whole all-round intelligent management system of breast cancer postoperative |
-
2021
- 2021-09-26 CN CN202111128954.7A patent/CN113871020B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU5355199A (en) * | 1996-09-30 | 1999-12-02 | Smithkline Beecham Corporation | Disease management method and system |
| CN109147943A (en) * | 2018-06-28 | 2019-01-04 | 刘毅 | System is assessed and instructed to perioperative |
| CN111383770A (en) * | 2020-03-09 | 2020-07-07 | 北京大学 | Clinical research and follow-up visit fusion system and method |
| CN113506630A (en) * | 2021-07-08 | 2021-10-15 | 上海中医药大学附属龙华医院 | Whole all-round intelligent management system of breast cancer postoperative |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113871020A (en) | 2021-12-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113871020B (en) | Health management method and system after critical illness diagnosis based on AI machine learning | |
| Slatter | Building expert systems: Cognitive emulation | |
| CN110111886A (en) | A kind of intelligent interrogation system and method based on XGBoost disease forecasting | |
| CN109213864A (en) | Criminal case anticipation system and its building and pre-judging method based on deep learning | |
| US20080222064A1 (en) | Processes and Systems for Automated Collective Intelligence | |
| CN107451230A (en) | A kind of answering method and question answering system | |
| CN109101624A (en) | Dialog process method, apparatus, electronic equipment and storage medium | |
| Bignold et al. | A conceptual framework for externally-influenced agents: An assisted reinforcement learning review | |
| US20210089965A1 (en) | Data Conversion/Symptom Scoring | |
| US20230316095A1 (en) | Systems and methods for automated scribes based on knowledge graphs of clinical information | |
| CN115827968A (en) | Individualized knowledge tracking method based on knowledge graph recommendation | |
| Bhatia et al. | Prediction of severity of diabetes mellitus using fuzzy cognitive maps | |
| CN117056536B (en) | Knowledge graph driving-based virtual doctor system and operation method thereof | |
| CN111754370B (en) | Artificial intelligence-based online education course management method and system | |
| CN117573946A (en) | Dialogue sample generation method, chat dialogue large model training method and related devices | |
| Hopkins et al. | Explaining the acquisition of a complex skill: Methodological and theoretical considerations uncovered in the study of simple addition and the moving-on process | |
| CN111783473A (en) | Method and device for identifying best answer in medical question and answer and computer equipment | |
| Kolonin et al. | Cognitive Architecture for Decision-Making Based on Brain Principles Programming | |
| Zhong | Structuralism? Functionalism? Behaviorism? Or mechanism? Looking for a better approach to AI | |
| CN117252260B (en) | Interview skill training method, equipment and medium based on large language model | |
| Laurikkala | Knowledge discovery for female urinary incontinence expert system | |
| US20240078266A1 (en) | Hierarchical tagging for personalized matching | |
| WO2023095069A1 (en) | An adaptive disease prediction system in an ai symptom-checker | |
| CN117936035A (en) | Psychological consultation comparison analysis method and system based on big data | |
| Yao | Reinforcement Learning for Healthcare: From Model Development to Deployment |
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 |