Early Detection of Mild Cognitive Impairment Utilizing Ocular Biomarkers-Based Risk Scoring Nomogram

Mohammed Seid Hussen¹Bess Yin-Hung Lam²Wei Gao³Liping Zhou⁴Kai Yip Choi^1*Henry Ho-lung Chan^1*

• ¹Laboratory of Experimental Optometry (Neuroscience), School of Optometry, The Hong Kong Polytechnic University, 11 Yuk Choi Road, Hung Hom, Kowloon, Hong Kong SAR 000000, China, Kowloon, Hong Kong, SAR China
• ²Department of Counselling and Psychology, Hong Kong Shue Yan University, 10 Wai Tsui Crescent, Braemar Hill, North Point, Hong Kong Island, Hong Kong SAR 000000, China, Hong Kong Island, Hong Kong, SAR China
• ³School of Computer Science, University of Technology, No 201 Fenghua Road, Ningbo 315211, China, Ningbo, China
• ⁴Research Centre for SHARP Vision (RCSV), The Hong Kong Polytechnic University, 11 Yuk Choi Road, Hung Hom, Kowloon, Hong Kong SAR 000000, China, Kowloon, Hong Kong, SAR China

Background The prevalence of cognitive impairment is increasing along with global aging. Early retinal structural and vascular changes, prior to the onset of clinically detectable retinal pathologies, have been increasingly associated with cognitive changes. However, the evidence related to the predictive performance of these biomarkers remains limited. Therefore, this study aimed to develop and validate a nomogram-based scoring tool for opportunistic screening of mild cognitive impairment (MCI). Methods This study prospectively recruited participants aged 60 years or older, including those with normal cognitive function. The retinal images were scanned using optical coherence tomography and angiography. Following the selection of potential predictors, a logistic regression model was built to predict MCI. Subsequently, a dynamic nomogram was developed to facilitate risk scoring in a clinical setting. The model's discriminative ability was evaluated using the area under the receiver operating characteristic curve, along with diagnostic metrics of sensitivity and specificity at 95% confidence interval (CI). The model was internally validated using bootstrapping. Decision curve analysis was conducted to evaluate the model's clinical impact and utility. Results The model indicated that central macular thickness (β: -1.13; 95% CI: -0.15,-2.15; p < 0.05), outer nasal perfusion density in the macular area (β:1.68; 95% CI: -2.92, -0.44; p = 0.008), and contrast sensitivity (β: -1.13; 95% CI: -2.03, -0.23; p < 0.05) were independently associated with MCI. This nomogram demonstrated a discriminative power of 0.90 (95% CI: 0.81, 0.98). The model also demonstrated good performance during bootstrap validation, achieving an AUC of 0.87. The optimal cutoff points achieved an accuracy of 86%, a sensitivity of 85% and a specificity of 87%. The decision curve analysis showed that the model provides a high net benefit. Conclusion This study developed and internally validated a dynamic, nomogram-based scoring tool for early detection of MCI that integrates non-invasive retinal and visual biomarkers. The model demonstrated high discriminative power and substantial clinical net benefit. Further evaluation of the model's prognostic value in predicting further cognitive decline may support its clinical utility.