Interpretable machine learning method for predicting multi-organ injury of diquat poisoning patient

CN121148684APending Publication Date: 2025-12-16NANJING DRUM TOWER HOSPITAL
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511245063.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-09-02
Publication Date
2025-12-16

AI Technical Summary

Technical Problem

Existing technologies are insufficient to capture early signs of multi-organ damage in patients with diquat poisoning in a timely and accurate manner, which may lead to delays in the optimal intervention time. As a result, the need to improve prognostic management and reduce the incidence and mortality of multi-organ failure remains unmet.

Method used

Interpretable machine learning methods were employed to construct XGBoost, CatBoost, and Random Forest models. Feature importance was evaluated through stepwise forward feature selection and SHAP values, generating a list of feature importance. The optimal algorithm and feature combination were then selected for multi-organ damage prediction.

Benefits of technology

It enables early and efficient prediction of multi-organ damage in patients with diquat poisoning, provides individualized intervention strategies, reduces the incidence of multi-organ failure and mortality, and the model is interpretable and stable, making it suitable for clinical diagnosis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121148684A_ABST
    Figure CN121148684A_ABST
Patent Text Reader

Abstract

The invention discloses an interpretable machine learning method for predicting multi-organ injuries of diquat poisoning patients, and the method comprises the steps: constructing and verifying a prediction system of eight kinds of organ injuries through a clinical database; toxic encephalopathy, myocardial injury, acute kidney injury, circulatory failure, skeletal muscle injury, liver injury, lung injury and gastrointestinal tract injury are covered. According to the method, early-stage efficient prediction of multi-organ injury is achieved, deep insight is provided for potential biological mechanisms of organ injury, clinical formulation of individualized intervention strategies is facilitated, the prognosis management level of diquat poisoning patients is expected to be remarkably improved, and the incidence rate and case fatality rate of multi-organ failure are reduced. According to the technical scheme, high prediction precision is kept, meanwhile, balance of interpretability, stability and clinical applicability of the model is achieved, a reliable auxiliary diagnosis tool is provided for organ injury risk assessment of acute poisoning patients, and the method has important clinical application value and wide market prospects.
Need to check novelty before this filing date? Find Prior Art