Randomized trial develops a machine learning risk model for cognitive frailty in older adults with Type 2 Diabetes Mellitus, suggesting personalized interventions may improve care.
Key Points
The aim is to develop a machine learning-based risk prediction model for cognitive frailty in older adults with type 2 diabetes mellitus.
349 participants recruited through convenience sampling
Randomly divided into a training set (n=244) and a test set (n=105)
Utilized six machine learning algorithms and SHAP for feature importance ranking.
23.5% of participants were identified with cognitive frailty
Support Vector Machine achieved highest AUC of 0.836 with accuracy of 0.759
Developed a web-based application for individual CF risk estimation.