Development and validation of a predictive nomogram for mild cognitive impairment in middle-aged and elderly populations: A multi-module analysis integrating air quality and sociodemographic factors

Abstract

Background

Growing evidence links air pollution to mild cognitive impairment (MCI), yet existing models often overlook environmental exposures. We developed a novel MCI nomogram integrating air pollution, sociodemographic, and clinical predictors for Chinese adults ≥ 45 years.

Objective

To develop and validate a personalized MCI risk assessment tool incorporating sociodemographic, clinical, and environmental factors.

Methods

Using the 2015 CHARLS cohort (n = 7702), we built two MCI prediction models with city- and county-level air pollution exposures. Model performance was assessed via discrimination (C-index, ROC), calibration, clinical utility (decision curves), predictive performance (net reclassification improvement, NRI and integrated discrimination improvement, IDI).

Results

This study analyzed 7702 participants, randomly split into training (n = 5391) and validation (n = 2311) groups. Uni-variate analysis identified MCI risk in age ≥75 years, private medical insurance, married individuals and males. Multivariate analysis in Model 2 identified 13 key factors associated with MCI, among them age ≥75 years (OR = 5.437, 95%CI: 3.524–8.388, p < 0.001), private medical insurance (OR = 4.994, 95%CI: 2.340–11.337, p = 0.002), being mental disorders (OR = 2.210, 95%CI: 1.217–4.014, p < 0.001), males (OR = 0.638, 95%CI: 0.493–0.824, p = 0.04), PM₁₀ (OR = 1.059, 95%CI: 1.051–1.067, p < 0.001) and PM_2.5 (OR = 1.025, 95%CI: 1.013 −1.038, p < 0.001) as strongest predictors. Model 2 outperformed Model 1 with higher discrimination, better calibration, greater clinical utility and improved accuracy.

Conclusions

The validated nomogram enables individualized MCI risk stratification, supporting targeted community prevention strategies.