Duke Health researchers have developed a machine learning model that can distinguish mild cognitive impairment from normal cognition using retinal images, offering a potential non-invasive and cost-effective method for early detection of cognitive decline. Published in Ophthalmology Science, this breakthrough could help identify individuals at risk of progressing to Alzheimer’s disease.
Senior author Sharon Fekrat, M.D., explained, “This work is exciting because we’ve struggled to differentiate mild cognitive impairment from normal cognition in previous models. It brings us closer to early detection before it escalates to Alzheimer’s dementia.”
The model builds on previous work that successfully identified Alzheimer’s patients through retinal scans. This new model leverages machine learning techniques to detect mild cognitive impairment, a precursor to Alzheimer’s, by identifying specific features in retinal images and considering patient data such as age, sex, visual acuity, and education years.
The research revealed a 79% sensitivity and 83% specificity in distinguishing individuals with mild cognitive impairment from those with normal cognition. This pioneering study is the first to use retinal images for this purpose and offers a non-invasive, cost-effective means to identify at-risk patients.
“The retina is a window to the brain, and machine learning can be a potent tool to screen patients at scale,” emphasized co-lead author Alexander Richardson, from the Eye Multimodal Imaging in Neurodegenerative Disease lab at Duke.
This study, supported in part by the Alzheimer’s Drug Discovery Foundation, is a significant step toward early cognitive impairment detection.
Source Duke Eye Center
