Interpretable machine learning for predicting mild cognitive impairment in elderly patients with type 2 diabetes mellitus: model development and performance assessment
Randomized trial develops a predictive model for mild cognitive impairment in elderly type 2 diabetes patients, implying improved screening strategies.
Key Points
Aim to develop and validate an interpretable machine learning model for predicting mild cognitive impairment in elderly patients with type 2 diabetes mellitus.
Analyzed a retrospective cohort of 923 elderly T2DM patients.
Data collected from January 2021 to January 2025.
Selected predictors using Boruta and LASSO feature selection, evaluated six ML algorithms.
RF model achieved an AUC of 0.842 in validation set, indicating strong predictive performance.
Duration of diabetes was identified as the most influential predictor, followed by HbA1c and FPG.
Overall MCI prevalence was 45.9%, with comparable rates in training and validation sets.