衰弱风险预测模型的范围综述.

陈红爽; 王琴潞; 邹海欧

Objective: To systematically integrate the construction and application status, predictors and performance of frailty risk prediction models in China and abroad. Methods：The Chinese and English databases were retrieved, and retrieval time from database establishment to June 30, 2022. Data was extracted and summarized for analysis. According to Prediction model Risk Of Bias Assessment Tool, the risk of bias was evaluated from four aspects with research objects, predictor, outcome and analysis of the models established. Results：A total of 12 literature was included in the study, which was mainly carried out by Chinese. The research subjects were mainly elderly people. The model construction methods could be divided into Logistic regression model and machine learning. The model mainly displayed the risk scoring formula based on regression coefficients of each factor. The three predictive factors with the highest frequency of occurrence were comorbidities, age, and multiple drug use. Conclusions：The models included in this study have good prediction, but the overall bias risk of the study is higher. In the future, visual model presentation methods can be applied, and models with low bias risk, good predictive performance, and high clinical practicality can be established.

CHINA; ONLINE information services; MEDICAL databases; CINAHL database; FRAIL elderly; PREDICTIVE tests; MEDICAL information storage & retrieval systems; SYSTEMATIC reviews; AGE distribution; POLYPHARMACY; MACHINE learning; RISK assessment; PREDICTION models; LOGISTIC regression analysis; MEDLINE; COMORBIDITY

Chinese Nursing Research, 2024, Vol 38, Issue 2, p293

1009-6493

Academic Journal

10.12102/j.issn.1009-6493.2024.02.016