Retinal imaging algorithm detects Alzheimer's

A study has shown that a deep learning algorithm using retinal imaging alone can detect potential risk of Alzheimer's disease with ‘good accuracy’.

A team, led by the Chinese University of Hong Kong, trained, validated and tested a deep learning proof-of-concept model using data from 12,949 retinal photographs, with 648 patients with Alzheimer's disease and 3,240 people without the disease. Results revealed 83.6% accuracy, 93.2% sensitivity, and 82% specificity for detecting Alzheimer's disease-dementia.

As Alzheimer's is a complex disease, which usually requires extensive neurological and behavioural tests, using a retinal imaging-based deep learning algorithm method would enable a 'simple, low-cost, low labour-dependent approach to identify patients in community settings with reasonable accuracy and sensitivity,’ according to the study.