Related Experiment Video
Updated: Jun 11, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Predictors of mild cognitive impairment in older adults living with HIV and multimorbidity: a comparative analysis
Chunxing Ge1, Huiying Gao2, Lanting Xia1
1School of Nursing and Rehabilitation, Nantong University, Nantong, China.
Background:
With the aging of the global population of people living with HIV (PLWH), cognitive impairment has emerged as an important public health concern. Older adults with HIV frequently experience multimorbidity, which may further increase the risk of mild cognitive impairment (MCI). However, studies exploring predictors of MCI among older adults living with HIV and multimorbidity remain limited.
Objective:
To investigate the prevalence and factors associated with screening-defined mild cognitive impairment (MCI) among older adults living with HIV and multimorbidity, and to explore the potential predictive value of logistic regression and decision tree models.
Methods:
A cross-sectional study was conducted among 327 older adults (aged ≥50 years) living with HIV and at least one comorbid chronic condition. Sociodemographic characteristics, clinical information, self-management ability, and social support were collected through structured questionnaires. Cognitive function was assessed using the Montreal Cognitive Assessment (MoCA), and dementia was excluded using the Mini-Mental State Examination (MMSE). Univariate analysis and multivariable logistic regression were performed to identify factors associated with MCI. A CHAID decision tree model with 10-fold cross-validation was constructed to explore hierarchical relationships among predictors. Receiver operating characteristic (ROC) curves were used to assess model performance.
Results:
Among the 327 participants, the prevalence of screening-defined MCI was 48.9%. Multivariable logistic regression analysis showed that older age, female sex, hypertension, monthly income of 3,001-5,999 RMB, and a higher number of comorbidities were significantly associated with an increased risk of cognitive impairment (p < 0.05). In contrast, higher education level, HIV knowledge learning experience, greater social support, and better daily living management ability were protective factors (p < 0.05). The decision tree model identified five key predictors, including number of comorbidities, education level, age, hypertension, and HIV knowledge learning experience, with number of comorbidities being the most important splitting variable. The area under the ROC curve (AUC) of the logistic regression model was 0.961, which was significantly higher than that of the decision tree model (0.916; p < 0.01).
Conclusion:
Screening-defined cognitive impairment is highly prevalent among older adults living with HIV and multimorbidity. Factors including the number of comorbidities, age, education level, HIV knowledge learning experience, and hypertension were identified as important correlates of cognitive impairment, with consistent findings across logistic regression and decision tree analyses. Both models demonstrated acceptable discriminatory ability within the study sample; however, these findings should be interpreted as exploratory given the lack of external validation. Overall, the results may contribute to the early identification of individuals at higher risk of cognitive impairment and provide a basis for developing targeted interventions. Further studies with rigorous validation are warranted to confirm the generalizability of these findings.
