Medical intelligent risk prediction method and system
By constructing and matching feature templates of medical record data, extracting and weighting the processing of risk factors, the problem of early monitoring of moderate metabolic acidosis is solved, more accurate risk prediction and early warning is achieved, and patient monitoring and intervention efficiency in ICU environment is improved. .
Patent Information
- Application Number
- CN202411180013.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-12-17
AI Technical Summary
In the prior art, the non-timedness and missed detection of data monitoring make it difficult to achieve early assisted decisions for moderate metabolic acidosis, especially in the ICU environment, where effective risk prediction and early warning mechanisms are lacking.
By extracting the characteristic parameter indicators in historical medical record information, building feature templates under topic type, performing medical record data summary processing and parameter matching, using weight value to weight processing, and verifying it with physiological parameters and medical record statement information, pushing warnings Information and monitoring suggestions, dynamically update weights to improve risk prediction accuracy.
It improves the prediction accuracy and early warning capabilities of the risk of moderate metabolic acidosis, ensures timely intervention and monitoring recommendations for patients in the ICU environment, and enhances decision-making support for medical staff.
Smart Images
Figure CN119153066A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of information processing, and in particular to a medical intelligent risk prediction method and system. Background Art
[0002] Metabolic acidosis refers to a type of acidosis that occurs when the body's pH drops due to metabolic factors. The normal human pH is between 7.35 and 7.45. If the patient's pH is less than 7.35, it means that acidosis may occur. Common metabolic acidosis include: excessive production of acidic substances; acidic substance excretion disorders; excessive loss of alkaline substances. There are many risk factors for metabolic acidosis. Usually, after metabolic acidosis occurs, intervention is usually to actively control blood sugar, rehydrate, correct shock, or give patients nutritional support. The common drug for metabolic acidosis is sodium bicarbonate, which is an alkaline drug that can correct acidosis.
[0003] In practice, due to the non-timeliness of data monitoring and the possibility of missed detection, that is, the non-exhaustiveness of monitoring, in order to better carry out auxiliary monitoring and prediction, the medical record database in the existing technology is integrated to provide an auxiliary decision-making method or system for early sodium bicarbonate administration in the ICU for moderate metabolic acidosis, and feature extraction of metabolic acid-related medical record data is performed to enhance auxiliary risk assessment, so as to carry out corresponding data monitoring and provide early warning information, which becomes a problem that needs to be solved. Summary of the invention
[0004] In order to solve one of the above problems, the present invention provides a medical intelligent risk prediction method, the method comprising:
[0005] Based on historical medical record information, characteristic parameter indicators related to moderate metabolic acidosis under different subject types are extracted, and characteristic templates under different subject types are constructed;
[0006] When obtaining the user's medical record data parameters, the medical record data is summarized and multiple subject types are extracted, and parameter matching under multiple subject types is performed for the same user;
[0007] Match the parameters of the same user under different subject types, weight the risk values of medical records with similar feature templates according to the weight values of different subject types, and push the weighted risk values to the user terminal;
[0008] Among them, before summarizing the medical records, it includes: obtaining the worst physiological parameters of the user within 12 hours, verifying the worst physiological parameter information with the user's main information, and when the verification warning exceeds the preset value, sending it to the corresponding expert group for data verification to correct the subject type.
[0009] Preferably, when obtaining the user's medical record data parameters, the medical record data is summarized and multiple topic types are extracted, including: identifying multiple risk factors related to moderate metabolic acidosis based on the user's medical record data; extracting a first keyword group from the multiple risk factors, the first keyword group including different topic types corresponding to multiple first keywords.
[0010] Preferably, the different topic types correspond to different diagnosis-related information.
[0011] Preferably, the physiological characteristic parameter indicators related to moderate metabolic acidosis include at least: blood pressure, PaO 2 、PaCO 2 , pH.
[0012] Preferably, verifying each worst physiological parameter information with the user's main information includes: obtaining the user's physiological parameters and extracting keywords in the medical record statement, comparing the standard reference information corresponding to the keywords in the case with the physiological parameters monitored by the user to determine whether they match, and performing alarm processing when the threshold is exceeded.
[0013] Preferably, matching is performed based on parameters under different subject types of the same user, weighted processing is performed on similar medical records based on weight values of different subject types, and the weighted risk values are pushed to the user terminal.
[0014] Preferably, the method further includes: performing statistical analysis on the results of historical medical records, obtaining the probability distribution of each risk factor in different subject scenarios, quantifying the interactions between multiple potential risk factors, configuring the relationship events between risk factors into a tree structure, and pushing monitoring suggestions to users based on the correlation between potential risk factors.
[0015] Preferably, the method further comprises: using Logistic multivariate regression analysis to screen out risk factors for metabolic acidosis in different subjects.
[0016] Preferably, the method further comprises: pushing monitoring suggestions to the user based on the correlation between potential risk factors, including pushing associated historical medical record information and monitoring items.
[0017] Preferably, a computer storage medium is provided, on which computer program instructions are stored for implementing the above method.
[0018] Preferably, a system is provided, the system comprising a memory and a processor, the memory storing a computer program, the computer program being executed by the processor to implement the above method.
[0019] The present invention discloses a medical intelligent risk prediction method and system, the method comprising: when obtaining the user's medical record data parameters, summarizing the medical record data, extracting multiple subject types, and matching parameters under multiple subject types for the same user; preprocessing the medical record data to correct the medical record subject type information, performing fitting through the corrected medical record subject types, and at the same time, proposing monitoring suggestions and dynamically updating weights through the historical medical record data of the risk factor association relationship constructed, thereby improving the accuracy of risk prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] The features and advantages of the present invention will be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the present invention in any way.
[0021] Figure 1 It is a schematic flow chart of the method of the present invention;
[0022] Figure 2 It is a schematic diagram of the application interface of the method of the present invention. DETAILED DESCRIPTION
[0023] These or other features and characteristics of the present invention, methods of operation, functions of related elements of the structure, combinations of parts, and economies of manufacture may be better understood with reference to the following description and drawings, which form a part of the specification. However, it is clearly understood that the drawings are used for illustration and description purposes only and are not intended to limit the scope of protection of the present invention. It is understood that the drawings are not drawn to scale. A variety of structural diagrams are used in the present invention to illustrate various variations according to embodiments of the present invention.
[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0025] It should be noted that the “ / ” in this article means or, for example, A / B can mean A or B; the “and / or” in this article is only a way to describe the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.
[0026] It should be noted that, in order to facilitate the clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, the words "first", "second", etc. are used to distinguish the same items or similar items with basically the same functions or effects. Those skilled in the art can understand that the words "first", "second", etc. do not limit the quantity and execution order. For example, the first information and the second information are used to distinguish different information, rather than to describe a specific order of information.
[0027] It should be noted that, in the embodiments of the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.
[0028] Example 1
[0029] like Figure 1 As shown, a medical intelligent risk prediction method is provided, the method comprising: extracting characteristic parameter indicators related to moderate metabolic acidosis under different subject types according to historical medical record information, and constructing characteristic templates under different subject types;
[0030] When obtaining the user's medical record data parameters, the medical record data is summarized and multiple subject types are extracted, and parameter matching under multiple subject types is performed for the same user;
[0031] Match the parameters of the same user under different subject types, weight the risk values of medical records with similar feature templates according to the weight values of different subject types, and push the weighted risk values to the user terminal;
[0032] Among them, before summarizing the medical records, it includes: obtaining the worst physiological parameters of the user within 12 hours, verifying the worst physiological parameter information with the user's main information, and when the verification warning exceeds the preset value, sending the corresponding expert to perform data verification to correct the subject type.
[0033] Exemplarily, the method includes constructing a correlation between different themes of different moderate metabolic acidosis cases and a set of physiological parameter index items; extracting characteristic parameter indicators from information in different cases, and constructing characteristic templates under different themes. When the data parameters of the patient to be tested are obtained, the data parameters are matched with different characteristic templates in the system for recommendation. In a preferred embodiment, the matching recommendation specific user terminal, i.e., the user, can delete and set different characteristic parameter indicators based on the information pushed by the characteristic template, and pre-set the characteristic parameters of the case information. The system extracts the medical record information of the patient, i.e., the user, and performs summary extraction on different themes.
[0034] Optionally, the characteristic parameter may be information of the worst physiological parameters filled in the pre-designed scale within the first 12 hours after the patient enters the emergency department; illustratively, the pre-designed scale at least includes: the criteria for determining metabolic acidosis: blood pH value <7.35 and carbon dioxide partial pressure (PaCO 2 )<45mmHg(1mmHg=0.133kPa).
[0035] The system obtains the characteristic parameters of users under different topics and matches the corresponding case characteristic template parameters, and weights the matching characteristic template cases according to the preset weight values of different templates under the same topic. The weighting is set as the score coefficient given by different experts in the historical matching. For example, different experts make weighted evaluations on various parameters of the cases pushed to the patients.
[0036] Among them, the weighted evaluation can also adaptively adjust the weight values of different historical medical record template parameters according to the patient's dynamic monitoring data. By updating the weight values, the evaluation indicators in the database are updated, and the system gives a comprehensive judgment result based on the medical record information of the corresponding feature samples.
[0037] Specifically, the matching may be to use the Euclidean distance algorithm to calculate the similarity between the feature values of the user's medical records and the risk feature values in historical cases and feature template cases; specifically, to standardize the feature values extracted from the user's medical records and historical cases.
[0038] For example, the eigenvalues can be normalized to the minimum or maximum, and the Euclidean distance calculation can be performed between the eigenvalues extracted from each user's plan and the eigenvalues of all historical cases, and the distance matrix between the user's medical record plan and all historical cases can be calculated. The minimum distance value corresponding to each user's medical record is taken, that is, the distance to the most similar historical case. The smaller the distance value, the more similar the two eigenvalue vectors are. The matching pairs in different distance intervals are counted to evaluate the overall matching degree. The matching parameters corresponding to the medical record plan are based on matching under different subject types.
[0039] Among them, the worst physiological parameter information is verified with the user's main information, including: keyword extraction of the patient's physiological parameter data and key information in the disease course description, mapping the main description and physiological parameter data indicators in the disease course, obtaining the standard template data in the system database, and verifying the standard template data with the parameter data obtained by monitoring to determine whether they match. If the number of mismatches exceeds the threshold, the corresponding patient information is sent to the associated expert group, and the subject information corresponding to the keyword is corrected.
[0040] Physiological parameters may include: blood pressure, Minimum PaO 2 , Maximum PaCO 2 , bicarbonate, Minimum pH, blood sugar, etc. Exemplarily, when the patient's medical history is considered to have a history of hypertension and diabetes, and when the physiological parameters within 12 hours of admission do not match the theme displayed by the above case keywords, an error prompt can be issued through a preset message template before the system template matches, and pushed to the attending physician or user interface. When the number of errors exceeds the preset threshold, the corresponding key theme message is pushed to the corresponding expert group according to the associated expert group pre-configured in the system by the attending physician, and the expert group updates the evaluation indicators, and feeds back the summary information of the revised main description, and then sends the revised patient summary information to the system for matching processing, matches the patient summary information with the feature template in the medical record database, and gives a comprehensive risk evaluation score based on the existing case database.
[0041] The above risk assessment score may be adapted to weighted processing of the medical record data in the history library based on the similarity of multiple case parameters. The weighted processing may be based on the comprehensive weighted score of the preceding cases according to the results of similar cases being pushed.
[0042] In an optional embodiment, due to the deviation of the time of detecting physiological parameters and the sampling samples, etc., and the risk monitoring of metabolic acidosis is a multi-cause and one-effect risk, the poisoning factors of metabolic acidosis are associated with multiple risk factors.
[0043] For this reason, when extracting feature templates, the jump model probability values of multiple risk factors can be constructed to provide risk warnings and risk prompts. In a preferred embodiment, the method can include extracting the jump probability values of risk factors under different themes based on historical medical record data cases, constructing risk jump probability values of different themes based on physiological indicators, and pushing monitoring suggestions to users based on the correlation relationship of potential risk factors, including pushing related historical medical record information and monitoring items.
[0044] In an optional manner, potential risk factors associated with moderate metabolic acidosis are determined based on historical medical record parameter data, and for each potential risk factor, its possible value range and distribution are determined, and parameters are set for each potential risk factor. In an optional manner, the value range and probability distribution of the parameter are determined based on historical data, expert evaluation, and database information. An exemplary probability distribution function is a standard normal distribution. The risk prediction value area parameters of metabolic acidosis with similar medical record monitoring courses are further given, and the risk prediction results are corrected in combination with the physiological parameters of the patient extracted at interval periods.
[0045] Among the optional methods: statistical analysis of the results of historical medical records, obtaining the probability distribution of each risk factor under different subject scenarios, quantifying the interaction between multiple potential risk factors, quantitative risk assessment and decision analysis, and configuring the relationship events between risk factors into a tree structure.
[0046] For example, for moderate metabolic acidosis, under the corresponding classification topic, the corresponding risk factors include a more than 3-fold decrease in the insulin secretion index, a more than 3-fold increase in the insulin resistance index, blood glucose>15.0mmo l / L, and a more than 3-fold increase in lactate.
[0047] In a preferred embodiment, the interaction relationship between risk factors can be constructed as a Bayesian network model: a binary node, i.e., a root node, is established in the Bayesian network for each event in the risk factor tree and named according to the event, and only one node is established for repeated events, i.e., repeated basic events in the risk factor tree represent only one root node in the Bayesian network;
[0048] Determine the prior probability distribution of the root node in the Bayesian network according to the failure distribution of the basic events in the risk factor tree; establish corresponding binary nodes for the logic gates in the risk factor tree, whose node flags and state values are consistent with the output of the logic gates in the risk tree, and name them according to the output events of the logic gates, and only establish one node for repeated output events; establish connections between nodes in the Bayesian network according to the connection relationship between the top events, intermediate events and basic events in the risk tree; determine the joint probability of occurrence of non-root nodes in the Bayesian network according to the logical relationship in the risk tree.
[0049] In an exemplary manner, the method identifies risk factors based on historical medical record data information, forms a risk list and a risk tree, and uses the user to edit risk events, risk factors, risk lists and risk trees, and estimates the risk probability based on the risk assessment method of the input medical record profile information, such as Figure 2 As shown, the weight of each risk factor can be automatically generated based on database reasoning to determine the risk level.
[0050] Exemplarily, Logistic multivariate regression analysis is used to screen out risk factors for metabolic acidosis under different themes, such as exemplarily extracting risk factors associated with patients with metabolic acidosis, such as smoking, drinking, chronic obstructive pulmonary disease, pneumonia, cirrhosis, diabetes, PaCO 2 , lactate, blood glucose, blood ketones, blood cortisol, endogenous creatinine clearance, serum creatinine, blood urea nitrogen, and blood potassium levels, PaO 2, FEV1 / FVC%, blood insulin, lactate dehydrogenase, urine volume and blood calcium levels decreased (all P<0.05), etc. The above risk factor information was associated with the medical records under different topics, and the theme templates under the parameters of different topics were matched with the information in the historical database to obtain the risk factors under different topics. The matching scores under each topic were weighted to finally obtain the predicted results.
[0051] For example, risk factors for metabolic acidosis after anticoagulation with sodium citrate include low PaO 2 , high PaCO 2 , high lactate, high blood sugar, high blood ketones, high blood potassium, high plasma creatinine, high blood urea nitrogen, liver cirrhosis, diabetes, intestinal resection and anastomosis. That is, the risk factor values under different topics are different. For this reason, the prediction results can be further corrected to correct the output, and the high-risk patients at admission can be comprehensively analyzed and evaluated, and the corresponding risks and related factors can be adjusted to facilitate the evaluation of the corresponding data.
[0052] Example 2
[0053] The present invention provides a computer storage medium, on which a computer program instruction is stored, for implementing the above method. The present invention provides a system, which includes a memory and a processor, on which a computer program is stored, and which is executed by the processor to implement the above method.
[0054] Those skilled in the art can understand that the implementation of all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.
[0055] As used in this application, the terms "component", "module", "system", etc. are intended to refer to a computer-related entity, which can be hardware, firmware, a combination of hardware and software, software, or software in operation. For example, a component can be, but is not limited to: a process running on a processor, a processor, an object, an executable file, a thread in execution, a program, and / or a computer. As an example, an application running on a computing device and the computing device can both be components. One or more components can exist in a process and / or thread in execution, and a component can be located in a computer and / or distributed between two or more computers. In addition, these components can be executed from various computer-readable media having various data structures thereon. These components can communicate in a local and / or remote process manner, such as according to a signal having one or more data packets (e.g., data from a component that interacts with another component in a local system, a distributed system, and / or interacts with other systems in a signal manner through a network such as the Internet).
[0056] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A medical intelligent risk prediction method, It is characterized in that The method comprises: Based on historical medical record information, characteristic parameter indicators related to moderate metabolic acidosis under different subject types are extracted, and characteristic templates under different subject types are constructed; When obtaining the user's medical record data parameters, the medical record data is summarized and multiple subject types are extracted, and parameter matching under multiple subject types is performed for the same user; Match the parameters of the same user under different subject types, weight the risk values of medical records with similar feature templates according to the weight values of different subject types, and push the weighted risk values to the user terminal; Among them, before summarizing the medical records, it includes: obtaining the user's worst physiological parameters within 12 hours, verifying the worst physiological parameter information with the user's main information, and when the verification warning exceeds the preset value, sending it to the corresponding expert group for data verification to correct the subject type.
2. The method according to claim 1, Its characteristics also include: When obtaining the user's medical record data parameters, the medical record data is summarized and processed to extract multiple topic types, including: identifying multiple risk factors related to moderate metabolic acidosis based on the user's medical record data; extracting a first keyword group from the multiple risk factors, the first keyword group including different topic types corresponding to multiple first keywords.
3. The method according to claim 2, Its characteristics also include: The different topic types correspond to different diagnosis-related information.
4. The method according to claim 3, Its characteristics also include: Physiological characteristic parameters associated with moderate metabolic acidosis include at least: blood pressure, PaO 2 、PaCO 2 , pH.
5. The method according to claim 4, Its characteristics also include: Verifying the worst physiological parameter information with the user's main information includes: obtaining the user's physiological parameters and extracting keywords in the medical record statement, comparing the standard reference information corresponding to the keywords in the main description in the case with the physiological parameters monitored by the user to determine whether they match, and issuing an alarm when the threshold is exceeded.
6. The method according to claim 5, Its characteristics also include: Matching is done based on the parameters of different topic types for the same user, weighted processing is performed on similar medical records based on the weight values of different topic types, and the weighted risk values are pushed to the user terminal.
7. The method according to claim 6, Its characteristics are also that The method further includes: performing statistical analysis on the results of historical medical records, obtaining the probability distribution of each risk factor under different subject scenarios, quantifying the interactions between multiple potential risk factors, configuring the relationship events between risk factors into a tree structure, and pushing monitoring suggestions to users based on the association between potential risk factors.
8. The method according to claim 7, Its characteristics also include: The method further includes: using Logistic multivariate regression analysis to screen out risk factors for metabolic acidosis under different subjects.
9. A computer storage medium having computer program instructions stored thereon for implementing any of the methods in claims 1-8.
10. A medical intelligent risk prediction system, the system comprising a memory and a processor, the memory storing a computer program, the computer program being executed by the processor to implement any one of the methods in claims 1-8.