Here’s something AI might yield a big change for us- a useful test for Alzheimer’s.  An affliction that affects some 7 million Americans.  And, we can detect its potential some 20 years before symptoms manifest.

Of course, that creates problems in my mind.  We have no treatment, we have no cure.  So knowing I will be subject to Alzhemer’s only increases my anxiety.  But, maybe there can be a benefit.

RetiSpec, a Toronto based startup, describes the process thusly.  They image the eyes using conventional equipment- but with a specialized (hyperspectral) camera. The camera captures rich images, containing a great deal of information, that is then analyzed by artificial intelligence. It compares those images to others that have biological signatures manifesting neurodegenerative diseases.

In particular, this approach is based on the hypothesis that beta amyloid (the peptide that accumulates in the brain of Alzheimer’s patients) accumulates in the retina.  Drs. Robert Vince and Swati Moore (U Minnesota) started this research in 2014.  They used this hyperspectral imaging (captures an image at various wavelengths) that can then identify the amyloid aggregates.  This concept was then licensed to RetiSpec.

A study of 108 subjects, at risk or manifested preclinical Alzheimer’s, were then subject to the hyperspectral imaging.  The results were compared to PET [Positron Emission Tomography] and CSF [Cererbrospinal Fluid] (those tests are very expensive and diagnosis takes weeks; moreover, insurance doesn’t cover the tests) data- and found the imaging was accurate 86% of the time.

It is the AI algorithm that provides the concept its power.  It can find patterns we can’t discern.  And, as more tests are preformed, AI will have larger databases to analyze, which will help with the results.

We await the next phase of clinical trials.

Share this: