AI Technology Can Potentially Diagnose Alzheimer's Early
The earlier Alzheimer's is diagnosed, obviously the better. Treatment tends to have a greater impact in the early stages of the disease. However, the problem has been arriving at an early diagnosis—it has historically posed challenges for doctors and researchers. The disease has been linked to metabolic changes; this makes it much harder to diagnose, as metabolic changes are subtler and difficult to pinpoint.
This is where a recent study conducted by the Big Data in Radiology (BDRAD/) research group, is seeking to change the game. Their approach involves using AI to uncover changes in the brain which could potentially predict Alzheimer's.
The researchers utilized data from over 2000 brain images of over one thousand Alzheimer's patients. The algorithm they utilized was trained to recognize metabolic patterns and consequently taught itself when/how such patterns were related to Alzheimer's. After working on the gathered data, the scientist then used the algorithm on 40 more images it'd never reviewed before. The results: the algorithm had a 100% success rate as far as detecting the disease at least six years before diagnosis.
The researchers in the group suggest that the testing was done was on a relatively small sample—they need larger studies to confirm the usefulness of this algorithm when it comes to more accurately predicting this disease in patients. The algorithm could have a broader purpose as well; one researcher suggests that it can be used to help with the work of radiologists in terms of early intervention with certain conditions.
The future of AI in medicine is certainly exciting, especially as it pertains to diagnosing and treating Alzheimer's. Another potential direction for algorithms could be looking for those abnormal protein clumps in the brain which are specific to Alzheimer's. The predictive power of AI will change how we approach and treat a broad spectrum of health conditions.