Early Identification of Mild Cognitive Impairment: An Innovative Model Using Ocular Biomarkers

Zhang Lingjing¹Yanwei Wang²Yuming Liu³Zi Ye^4,5*Zhaohui Li^5*

• ¹Nankai University, Tianjin, China
• ²Department of General Surgery, The 926th Hospital, Joint Logistics Support Force of PLA, Kaiyuan, China
• ³People's Liberation Army General Hospital, Beijing, China
• ⁴Department of Ophthalmology, People's Liberation Army General Hospital, Haidian, Beijing, China
• ⁵Department of Ophthalmology, The General Hospital of the People's Liberation Army, Beijing, China

Background: Alzheimer's disease (AD) is a neurodegenerative disorder characterized by progressive, irreversible brain damage. Current diagnostic procedures for AD are both costly and highly invasive for patients. Age-related cataract (ARC), a common ocular condition in elderly populations, correlates with a 1.43-fold increased risk of developing AD. This study sought to establish a novel model for early detection of mild cognitive impairment (MCI) in patients with ARC.The study prospectively collected 170 monocular data as training dataset and 65 monocular data from another independent medical center as test dataset.Demographic data and comprehensive ophthalmic examination results were collected.The least absolute shrinkage and selection operator (LASSO) method and multivariate logistic regression analysis were performed using R software for dimensionality reduction and variable selection. A nomogram was constructed, and its discriminative ability was evaluated using receiver operating characteristic (ROC) curve, area under the ROC curve (AUC) with 95% confidence interval (CI), as well as sensitivity and specificity. Internal validation was performed using 1000-resample bootstrap analysis, while model calibration was assessed through calibration curves and Brier scores.Decision curve analysis (DCA) was performed to evaluate clinical utility. A baseline model incorporating demographic variables was developed for comparison with the nomogram. Additionally, an external dataset from an independent medical center was employed as a test set to further validate the nomogram's predictive performance. An online calculator was created using the "DynNom" and "rsconnect" functions.Results: Through LASSO regression and multivariate logistic regression analyses, six variables were identified and incorporated into the nomogram: age (OR: 1.097; 95%CI:1.041-1.161; p < 0.001), years of education (OR: 0.333; 95%CI: 0.