CN-122025153-A - Clinical intervention decision-making method and device, electronic equipment and storage medium

CN122025153ACN 122025153 ACN122025153 ACN 122025153ACN-122025153-A

Abstract

The embodiment of the application provides a clinical intervention decision-making method and device, electronic equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the steps of obtaining secretory otitis media clinical data of a user to be evaluated, obtaining a target secretory otitis media disease probability prediction model, carrying out feature evaluation on the clinical features of the multiple etiology agents based on the target secretory otitis media disease probability prediction model to obtain a plurality of user feature scores, carrying out disease probability prediction on the user to be evaluated based on the target secretory otitis media disease probability prediction model and the plurality of user feature scores to obtain user disease prediction data, and carrying out clinical intervention decision on the user to be evaluated based on the user disease prediction data. The implementation of the application can effectively prevent the occurrence of secretory otitis media and improve the accuracy of clinical intervention decisions, and can effectively evaluate the treatment effect of the secretory otitis media.

Inventors

LIAO ZHIFANG
JIANG HONGYAN
KE CHAOYANG
TU BO
ZHANG LUMEI
CHEN JING

Assignees

深圳市人民医院

Dates

Publication Date: 20260512
Application Date: 20251113

Claims (10)

1. A method of clinical intervention decision making, the method comprising: acquiring secretory otitis media clinical data of a user to be evaluated, wherein the secretory otitis media clinical data comprises a plurality of etiology clinical features; acquiring a target secretory otitis media disease probability prediction model; Based on the target secretory otitis media disease probability prediction model, performing feature evaluation on the clinical features of the multiple causes to obtain multiple user feature scores; Based on the target secretory otitis media illness probability prediction model and the plurality of user feature scores, carrying out illness probability prediction on the user to be evaluated to obtain user illness prediction data; and carrying out clinical intervention decision on the user to be evaluated based on the user illness prediction data.
2. The method according to claim 1, wherein the training method for obtaining the target secretory otitis media disease probability prediction model in the target secretory otitis media disease probability prediction model comprises: acquiring secretory otitis media detection sample data, wherein the secretory otitis media detection sample data comprises a result tag and a plurality of characteristic tags; acquiring a likelihood function and a probability transformation function of secretory otitis media; Based on the result label and the plurality of characteristic labels, performing weight fitting on the secretory otitis media likelihood function to obtain characteristic weights; based on the feature weights, performing function adjustment on the likelihood function of the secretory otitis media to obtain feature scoring function of the secretory otitis media; and performing function combination on the characteristic scoring function of the secretory otitis media and the probability transformation function to obtain the target secretory otitis media disease probability prediction model.
3. The method of claim 2, wherein the target secretory otitis media disease probability prediction model comprises the secretory otitis media feature scoring function comprising the feature weights, wherein the feature evaluating the plurality of etiological clinical features based on the target secretory otitis media disease probability prediction model results in a plurality of user feature scores comprising: constructing a feature score mapping table based on the feature weights and the plurality of feature labels; And carrying out score searching on the feature score mapping table based on the clinical features of the multiple etiologies to obtain the feature scores of the multiple users.
4. The method of claim 3, wherein constructing a feature score mapping table based on the feature weights and the plurality of feature labels comprises: determining a plurality of factor secretory otitis media risk characteristics based on the plurality of characteristic tags; based on the feature weights, feature screening is carried out on risk features of the secretory otitis media of various factors to obtain the feature with the largest weight; Giving the score to the maximum weight characteristic to obtain a reference score; Based on the reference score and the feature weight, performing score assessment on risk features of the secretory otitis media of the multiple factors to obtain feature scores; And constructing a mapping table based on the feature scores and the risk features of the secretory otitis media of various factors to obtain the feature score mapping table.
5. The method of claim 2, wherein the target secretory otitis media disease probability prediction model includes the secretory otitis media feature scoring function and the probability transformation function, wherein the predicting the disease probability of the user to be evaluated based on the target secretory otitis media disease probability prediction model and the plurality of user feature scores to obtain user disease prediction data includes: Based on the feature scoring function of the secretory otitis media, performing total score calculation on the feature scores of the plurality of users to obtain feature total scores of the users; based on the probability conversion function, carrying out illness probability conversion on the user characteristic total score to obtain user illness probability; performing level quantification on the user illness probability to obtain a user illness risk level; and combining information of the user illness risk level and a preset risk level decision text to obtain the user illness prediction data.
6. The method of any one of claims 1-5, wherein the user disease prediction data comprises follow-up observation prediction data, and wherein the making of a clinical intervention decision for the user to be evaluated based on the user disease prediction data comprises: Performing high-score feature follow-up on the user to be evaluated based on the follow-up observation prediction data to obtain follow-up information, wherein the follow-up information comprises follow-up feature risk unchanged information or follow-up feature risk improvement information; generating non-clinical intervention decision information based on the follow-up feature risk invariant information; And generating clinical intervention decision-making information based on the follow-up characteristic risk improvement information.
7. The method of claim 6, wherein after making clinical intervention decisions for the user to be evaluated based on the user disease prediction data, the method further comprises: Based on the clinical intervention decision information, performing intervention on the user to be evaluated to obtain intervention effect monitoring information, wherein the intervention effect monitoring information comprises intervention effective information or intervention ineffective information; based on the intervention effective information, carrying out feature follow-up evaluation on the user to be evaluated to obtain user feature follow-up evaluation data; and performing secondary intervention on the user to be evaluated based on the intervention invalidation information to obtain secondary intervention effect monitoring information.
8. A clinical intervention decision making device, the device comprising: A data acquisition module for acquiring secretory otitis media clinical data of a user to be evaluated, wherein the secretory otitis media clinical data comprises a plurality of etiology clinical features; The model construction module is used for acquiring a target secretory otitis media disease probability prediction model; The feature evaluation module is used for performing feature evaluation on the clinical features of the multiple causes based on the target secretory otitis media disease probability prediction model to obtain multiple user feature scores; The illness probability prediction module is used for predicting the illness probability of the user to be evaluated based on the target secretory otitis media illness probability prediction model and the user feature scores to obtain user illness prediction data; And the clinical intervention decision module is used for carrying out clinical intervention decision on the user to be evaluated based on the user disease prediction data.
9. An electronic device comprising a memory storing a computer program and a processor implementing the clinical intervention decision method of any of claims 1 to 7 when the computer program is executed by the processor.
10. A computer readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the clinical intervention decision-making method of any of claims 1 to 7.

Description

Clinical intervention decision-making method and device, electronic equipment and storage medium Technical Field The present application relates to the field of artificial intelligence technologies, and in particular, to a method and apparatus for clinical intervention decision making, an electronic device, and a storage medium. Background The clinical intervention decision is used for judging whether to take clinical intervention measures according to clinical data of the user, for example, in a secretory otitis media diagnosis scene, by carrying out clinical intervention decision on the secretory otitis media diagnosis user, the treatment safety of the secretory otitis media diagnosis user can be ensured, the cure rate is improved, and the probability of the secretory otitis media diagnosis user causing the problem of ears due to the secretory otitis media is reduced. At present, most of common clinical intervention decision methods of the secretory otitis media are that medical staff analyze clinical data of users who visit the secretory otitis media according to own experience to obtain clinical data analysis information, further, whether clinical intervention measures are adopted for users who visit the secretory otitis media is judged according to the clinical data analysis information, but different clinical intervention decisions are easy to appear due to experience gaps of the medical staff, and besides, the clinical intervention decisions provided by the medical staff are not accurate enough, so that how to improve the accuracy of the clinical intervention decisions becomes a technical problem to be solved urgently. Disclosure of Invention The embodiment of the application mainly aims to provide a method and a device for clinical intervention decision-making, electronic equipment and a storage medium, aiming to improve the accuracy of clinical intervention decision-making. To achieve the above object, a first aspect of an embodiment of the present application proposes a clinical intervention decision-making method, the method comprising: Acquiring clinical data related to secretory otitis media of a user to be evaluated, wherein the clinical data related to secretory otitis media comprises a plurality of etiology clinical features; acquiring a target secretory otitis media disease probability prediction model; Based on the target secretory otitis media disease probability prediction model, performing feature evaluation on the clinical features of the multiple causes to obtain multiple user feature scores; Based on the target secretory otitis media illness probability prediction model and the plurality of user feature scores, carrying out illness probability prediction on the user to be evaluated to obtain user illness prediction data; and carrying out clinical intervention decision on the user to be evaluated based on the user illness prediction data. In some embodiments, the training method for obtaining the target secretory otitis media disease probability prediction model comprises the following steps: acquiring secretory otitis media detection sample data, wherein the secretory otitis media detection sample data comprises a result tag and a plurality of characteristic tags; acquiring a likelihood function and a probability transformation function of secretory otitis media; Based on the result label and the plurality of characteristic labels, performing weight fitting on the secretory otitis media likelihood function to obtain characteristic weights; based on the feature weights, performing function adjustment on the likelihood function of the secretory otitis media to obtain feature scoring function of the secretory otitis media; and performing function combination on the characteristic scoring function of the secretory otitis media and the probability transformation function to obtain the target secretory otitis media disease probability prediction model. In some embodiments, the target secretory otitis media disease probability prediction model includes the secretory otitis media feature scoring function, the secretory otitis media feature scoring function includes the feature weights, and feature assessment is performed on the multiple etiology clinical features based on the target secretory otitis media disease probability prediction model to obtain multiple user feature scores, including: constructing a feature score mapping table based on the feature weights and the plurality of feature labels; And carrying out score searching on the feature score mapping table based on the clinical features of the multiple etiologies to obtain the feature scores of the multiple users. In some embodiments, the constructing a feature score mapping table based on the feature weights and the plurality of feature labels includes: Determining a plurality of secretory otitis media risk characteristics based on the plurality of characteristic tags; based on the feature weights, feature screening is carried out