Development and validation of Galectin-3 and CVAI-based model for predicting cognitive impairment in type 2 diabetes.

Abstract

Objective: The objective of this study is to develop a predictive model combining multiple indicators to quantify the risk of mild cognitive impairment (MCI) in T2DM patients.

Methods: This study included Chinese T2DM patients who were hospitalized at Zhongda Hospital between November 2021 and May 2023. Clinical data, including demographics, medical history, biochemical tests, and cognitive status, were collected. Cognitive assessment was performed using neuropsychological tests, and MCI was diagnosed based on the Montreal Cognitive Assessment (MoCA) scores. The dataset was randomly divided into a training set and a validation set in a 7:3 ratio. Logistic regression analysis was conducted to identify factors influencing MCI in the training set. A nomogram-based scoring model was then developed by integrating these findings with high-risk clinical variables, and its performance was validated in the validation set.

Results: In this study, T2DM patients were divided into a training set and a validation set in a 7:3 ratio. There were no significant differences in MCI incidence, demographics, or clinical characteristics between the two groups, confirming the appropriateness of model construction. In the training set, Galectin-3 and CVAI were significantly negatively correlated with cognitive function (MoCA and MMSE scores), and this negative correlation remained after adjusting for confounding variables. Logistic regression analysis revealed that age, CVAI, and Galectin-3 significantly increased the risk of MCI, while years of education had a protective effect. The constructed nomogram model, which integrated age, sex, education level, hypertension, CVAI, and Galectin-3 levels, exhibited high predictive performance (C-index of 0.816), with AUCs of 0.816 in the training set and 0.858 in the validation set, outperforming single indicators. PR curve analysis further validated the superiority of the nomogram model.

Conclusion: The straightforward, highly accurate, and interactive nomogram model developed in this study facilitate the early risk prediction of MCI in individuals with T2DM by incorporating Galectin-3, CVAI, and other common clinical risk factors.