Prediction model of daily living ability of stroke patients based on machine learning:a systematic review
WANG Fengting
SONG Jingwen
YU Yajing
LIN Hua
Abstract：Objective:To systematically evaluate prediction models for activities of daily living(ADL)in stroke patients based on machine learning.Methods:Literature on ADL in stroke patients was retrieved from PubMed,EMbase,Web of Science,CINAHL,ASA,ESO,CNKI,VIP,WanFang Data,and CBM.The retrieval time was from the inception to August 30,2024.Two reviewers independently screened,extracted data,and assessed bias.Results:A total of 9 studies were included,involving 10 algorithms and 26 prediction models.The AUC of the prediction models ranged from 0.74 to 0.98.Conclusion:Current evidence shows that the overall risk of bias in ADL prediction models for stroke patients is high,and the predictive performance is average.Models constructed using support vector machine(SVM)methods have achieved the best predictive performance in different studies.Age,Brunnstrom staging training,and Barthel index are common predictors factors in ADL prediction models for stroke patients.
Keywords：strokeactivities of daily livingmachine learningprediction modelsystem evaluationevidence-based nursing
Publication Date：2025-12-10
Online Publishing Date：2025-12-19(First online date of this platform, not the publication date of the document)
Pages：6( 4826-4831 )