Nomogram for predicting mild cognitive impairment in Chinese elder CSVD patients based on Boruta algorithm.

Abstract

Background: The number of patients with cerebral small vessel disease is increasing, especially among the elderly population. With the continuous improvement of detection techniques, the positivity rate keeps increasing. Our goal is to develop a nomogram for early identification of PSCI and PSCID in stroke patients.

Methods: In a retrospective cohort, chained data imputation was performed to ensure no statistical differences from the original dataset. Subsequently, Boruta algorithm was utilized for variable selection based on their importance, followed by logistic regression employing backward stepwise regression. Finally, the regression results were visualized as a Nomogram.

Results: The nomogram chart in this study achieves clinical utility in a concise and user-friendly manner, passing the Hosmer-Lemeshow goodness-of-fit test. ROC and calibration curves indicate its high discriminative ability.

Conclusion: While CSVD is prevalent among middle-aged and older individuals, cognitive decline trajectories differ. Endocrine metabolic indicators like IGF-1 offer early predictive value. This study has produced a succinct nomogram integrating demographic and clinical indicators for medical application.