CN-122024993-A - Medical follow-up management method and system based on large model prompt word engineering

CN122024993ACN 122024993 ACN122024993 ACN 122024993ACN-122024993-A

Abstract

The invention provides a medical follow-up management method and system based on large model prompt word engineering, and relates to the technical field of medical management. Aiming at the problem of poor pertinence of a follow-up strategy in the prior art, the method comprises the steps of obtaining a prompt word in a prompt word library according to a follow-up scene, optimizing the prompt word to obtain a final prompt word, collecting health state data and life habit data of a user, inputting the health state data and the life habit data into a follow-up dialogue model, generating a preset follow-up plan according to the input data, adjusting the preset follow-up plan according to the final prompt word by the follow-up dialogue model, outputting the final follow-up plan, obtaining behavior data related to execution of the final follow-up plan by the user, and updating the final prompt word in real time according to the health state data, the life habit data and the behavior data. The follow-up strategy generated by the medical follow-up management method provided by the invention has higher pertinence.

Inventors

YAN MING
WANG RUI

Assignees

天津新康医疗健康新技术科技发展有限公司
天津心悦医学影像诊断有限公司

Dates

Publication Date: 20260512
Application Date: 20260408

Claims (8)

1. A medical follow-up management method based on large model prompt word engineering is characterized by comprising the following steps: s1, acquiring a prompt word from a prompt word library according to a follow-up scene, and optimizing the prompt word through an open source model, wherein a small-parameter open source NLP model is adopted for diagnosing problems of the prompt word, and the small-parameter open source NLP model is obtained by fine tuning training based on a medical field labeling data set; aiming at the problem diagnosis result, automatically adjusting the content of the prompt words through the small-parameter open-source NLP model, wherein the method comprises the steps of respectively testing the prompt words before adjustment and after adjustment, and determining the final prompt words through a sampling algorithm; S2, acquiring health state data and life habit data of a user; S3, inputting the health state data and the life habit data into a follow-up dialogue model, and generating a preset follow-up plan according to the input data by the follow-up dialogue model; Specifically, a dynamic decision algorithm is embedded in the follow-up dialogue model, and the dynamic decision algorithm fuses input data to predict follow-up resource requirements of a user, wherein the follow-up resource requirements comprise follow-up frequency requirements, follow-up mode requirements and follow-up duration requirements; the follow-up dialogue model combines the follow-up resource requirements to generate a preset follow-up plan containing a resource allocation scheme; s4, the follow-up dialogue model adjusts a preset follow-up plan according to the final prompt word, and outputs a final follow-up plan; s5, acquiring behavior data related to the execution of the final follow-up plan by the user; s6, updating the final prompt word in real time according to the health state data, the life habit data and the behavior data.
2. The medical follow-up management method based on big model prompt word engineering according to claim 1, wherein the dynamic decision algorithm fuses input data to predict follow-up resource requirements of a user, and comprises the following steps: carrying out standardized processing on input data; according to the influence degree of the input data on the follow-up requirement, giving corresponding weight to the input data; Digging association features among input data through a machine learning model, and establishing a prediction model of multidimensional features and follow-up resource requirements; and outputting the follow-up resource requirement through the prediction model.
3. The medical follow-up management method based on big model prompt word engineering according to claim 1, wherein the follow-up scene comprises a disease type, the step of obtaining the prompt word in the prompt word stock according to the follow-up scene comprises the step of obtaining the prompt word associated with the disease type, and the step of adjusting a preset follow-up plan according to the final prompt word by the follow-up dialogue model in S4 comprises the step of supplementing follow-up contents of the disease type in the preset follow-up plan.
4. The medical follow-up management method based on big model prompt word engineering according to claim 3, wherein the follow-up scene further comprises a patient type, the step of obtaining the prompt word in the prompt word bank according to the follow-up scene comprises the step of obtaining the prompt word associated with the patient type, the prompt word comprises a speaking type, and the step of S4, the follow-up dialogue model adjusts a preset follow-up plan according to the final prompt word comprises the step of adjusting the speaking type in the preset follow-up plan.
5. The medical follow-up management method based on big model prompt word engineering according to claim 1, wherein in S5, obtaining behavior data related to execution of a final follow-up plan by a user includes: acquiring a feedback state of a user on a follow-up plan; and acquiring the updated state of the health state data.
6. The medical follow-up management method based on big model prompt word engineering according to claim 1, wherein in S6, the final prompt word is updated in real time according to the health status data, including: when the health state data show that the fluctuation range of the health index of the user exceeds a preset range, adding a conversation related to health risk early warning in the final prompt word; And synchronously storing the updated final prompting words to a prompting word stock.
7. The medical follow-up management method based on big model prompt word engineering according to claim 5, wherein in S6, the final prompt word is updated in real time according to the behavior data, including: when the behavior data show that the feedback state of the user on the follow-up plan does not meet the preset requirement, adding integral incentive guiding related content in the final prompt word; When the behavior data show that the feedback state of the user on the follow-up plan meets the preset requirement, adding positive relevant contents of health management effects in the final prompt word; And synchronously storing the updated final prompting words to a prompting word stock.
8. A medical follow-up management system based on big model prompt word engineering, characterized in that the system is used for executing the medical follow-up management method based on big model prompt word engineering according to any one of claims 1-7, the system comprises: The prompt word acquisition module is used for acquiring the prompt words in the prompt word library according to the follow-up scene, and optimizing the prompt words to obtain final prompt words; the acquisition module is used for acquiring health state data and life habit data of the user; The follow-up visit plan generation module is connected with the acquisition module and used for inputting the health state data and the life habit data into a follow-up visit dialogue model, and the follow-up visit dialogue model generates a preset follow-up visit plan according to the input data; the follow-up visit plan adjusting module is connected with the prompt word obtaining module and the follow-up visit plan generating module and is used for controlling the follow-up visit dialogue model to adjust a preset follow-up visit plan according to the final prompt word and outputting a final follow-up visit plan; the behavior data acquisition module is used for acquiring behavior data related to the execution of the final follow-up plan by the user; And the real-time updating module is connected with the behavior data acquisition module, the acquisition module and the prompting word acquisition module and is used for updating the final prompting word in real time according to the health state data, the life habit data and the behavior data.

Description

Medical follow-up management method and system based on large model prompt word engineering Technical Field The invention relates to the technical field of medical management, in particular to a medical follow-up management method and system based on large model prompt word engineering. Background Medical follow-up refers to the medical management behavior of a medical institution to continuously track the health state of a patient by periodically inquiring, data acquisition, health guidance and the like after the patient is discharged or after the staged treatment is completed. Potential health risks can be found in time by dynamically monitoring the postoperative recovery condition, the chronic disease control effect and the treatment compliance of the patient, and the readmission rate and the complication occurrence rate are reduced. The patent with publication number CN20220933A discloses an intelligent follow-up method based on a large model, which comprises the steps of obtaining follow-up data of a patient, and generating the follow-up problem of the patient aiming at the patient disease type by guiding the large model by adopting an instruction prompt word based on the preset follow-up problem corresponding to the patient disease type. The disclosed patent only collects the health state data of the patient, and does not consider life habit data and behavior feedback data of the patient to the follow-up procedure. Moreover, the prompt words cannot be updated in real time according to the health state data, life habit data and behavior data of the patient. In the prior art, life habit data and behavior feedback data of a patient on a follow-up flow are not considered, and meanwhile, prompt words cannot be updated in real time according to health state data, life habit data and behavior data of the patient, so that the follow-up strategy is difficult to adjust in a targeted manner. Therefore, the medical follow-up management method and system based on the large model prompt word engineering are developed, and have important significance for improving pertinence of follow-up strategies. Disclosure of Invention Aiming at the problem of poor pertinence of follow-up strategies in the prior art, the invention provides a medical follow-up management method based on large-model prompt word engineering, which specifically comprises the following steps: s1, acquiring a prompt word from a prompt word library according to a follow-up scene, and optimizing the prompt word through an open source model, wherein a small-parameter open source NLP model is adopted for diagnosing problems of the prompt word, and the small-parameter open source NLP model is obtained by fine tuning training based on a medical field labeling data set; aiming at the problem diagnosis result, automatically adjusting the content of the prompt words through the small-parameter open-source NLP model, wherein the method comprises the steps of respectively testing the prompt words before adjustment and after adjustment, and determining the final prompt words through a sampling algorithm; S2, acquiring health state data and life habit data of a user; S3, inputting the health state data and the life habit data into a follow-up dialogue model, and generating a preset follow-up plan according to the input data by the follow-up dialogue model; Specifically, a dynamic decision algorithm is embedded in the follow-up dialogue model, and the dynamic decision algorithm fuses input data to predict follow-up resource requirements of a user, wherein the follow-up resource requirements comprise follow-up frequency requirements, follow-up mode requirements and follow-up duration requirements; the follow-up dialogue model combines the follow-up resource requirements to generate a preset follow-up plan containing a resource allocation scheme; s4, the follow-up dialogue model adjusts a preset follow-up plan according to the final prompt word, and outputs a final follow-up plan; s5, acquiring behavior data related to the execution of the final follow-up plan by the user; s6, updating the final prompt word in real time according to the health state data, the life habit data and the behavior data. The dynamic decision algorithm is used for predicting follow-up resource demands of users by fusing input data, and comprises the steps of carrying out standardized processing on the input data, giving corresponding weights to the input data according to the influence degree of the input data on the follow-up resource demands, excavating association features among the input data through a machine learning model, establishing a prediction model of multidimensional features and the follow-up resource demands, and outputting the follow-up resource demands through the prediction model. Further, after the prompt words are obtained from the prompt word library according to the follow-up scene, the method further comprises the steps of optimizing the prompt words, specifically, pe