ARTIFICIAL INTELLIGENCE CAN HEAR EARLY ALZHEIMER’S ONSET

Although some progress has been made, Alzheimer’s Disease has always been difficult to diagnose with certainty. The most accurate tests tend to be expensive and health insurance policies don’t always cover them. 
Now researchers at McCann Healthcare Worldwide Japan and Kyoto University seem to have found something simpler and cheaper: listening to people as they speak.
Past research has shown that in the early stages of the illness, people begin to lose their specific words and substitute generic ones or placeholders. “Hand me the scissors” becomes “hand me the thing.”
Those changes can be too inconsistent or subtle for others to hear or pay attention to.
The Japanese study recorded more than 1,500 telephone conversations with 24 patients diagnosed as being in the early stages of Alzheimer’s and 99 people judged dementia-free.
Each participant took a standard cognitive test administered over the phone.
The scientists then fed the recordings into an artificial intelligence program and told it to sort the audios from Alzheimer’s patients from those who were cognitively healthy.
The AI returned the results with perfect accuracy: no cognitively healthy person was diagnosed as having Alzheimer’s and no Alzheimer’s patient was misdiagnosed as being cognitively healthy.
Now that the study team has created an AI program with the basic skills to diagnose Alzheimer’s by speech, the group will set about refining the program so it can make equally accurate diagnoses among persons with more subtle speech symptoms.
TRENDPOST: Alzheimer’s Disease has been shown to be halted, or even reversed, by lifestyle changes, as we reported in “New Breakthroughs May Cure Alzheimer’s” (13 Mar 2019). The earlier the diagnosis can be made, the sooner changes can be made in a person’s diet, nutrition, physical activity, and other patterns that can combat the illness.
By mid-century, Alzheimer’s will be widely recognized as a reversible condition.