Video Experimental Relacionado
Updated: Jan 13, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Desarrollo y validación de un modelo de predicción de riesgo de deterioro cognitivo leve en adultos chinos mayores
Lulu Yan1,2, Yuanyuan Peng3, Chenjiao Guo3
1Health Science Center, Yangtze University, Jingzhou City, Hubei Province, 434023, China.
Background:
As the population continues to age, the prevalence of mild cognitive impairment (MCI) has increased steadily. Studies have shown that older adults with chronic diseases are more likely to develop MCI than those without chronic conditions, suggesting that chronic diseases may play a significant role in the onset of MCI. Therefore, this study is designed to develop a predictive model for MCI among older individuals with chronic diseases in China and to identify the major factors influencing the occurrence of MCI.
Method:
The training and internal validation data are from the 2018 China Health and Retirement Longitudinal Study (CHARLS), using retrospective data with 4,712 older adults with chronic diseases. The external validation data are from the General Hospital of Southern Theater Command in Guangdong, using prospective data with 1,000 cases. This is an observational study. Univariate logistic regression was used to select statistically significant predictors, and ultimately, a combination of LASSO regression and random forest results identified 9 optimal predictors, which were used to construct the nomogram. The model's discrimination, calibration, clinical applicability, and generalizability were assessed using the receiver operating characteristic (ROC) curve, calibration curve, decision curve analysis (DCA), and internal validation.
Results:
Age, education level, child satisfaction, marital status, depressive symptoms, ADL score, income, SCD, and the number of chronic diseases were identified as significant predictors of MCI in older adults with chronic diseases. The AUC values exceeded 0.7 across the training, internal validation, and external validation sets.The calibration curves closely align with the diagonal, and the P values of the Hosmer-Lemeshow test are all greater than 0.05, indicating strong consistency between the predicted and actual outcomes. The DCA further demonstrates that the model has significant clinical utility.
Conclusion:
The nomogram prediction model deeloped in this study demonstrated good predictive performance and may serve as a useful tool to help identify older adults with chronic diseases who are at increased risk of MCI. These findings may inform future strategies for individualized risk assessment and early management.
