Health insurer calls analysed for signs of disease in your voice | New  Scientist

A few months ago we pointed to the discovery made by a group of German scientists who had managed to deactivate noisy neurons that prevented the retention of memories by people suffering from Alzheimer's.

Now, a group of researchers has been tasked with developing an AI-based tool that could help detect the disease in its early stages. This, through the analysis, carried out in the language of the sick.

This is how research by the Stevens Institute of Technology in New Jersey took as an object of study the way people with Alzheimer's are expressed.

Usually, with the advent of this disease, the parts of the brain involved in language are affected, causing the person to experience difficulties in using the right words when expressing themselves, changing the nouns to pronouns as an example.

Voice Analysis to Detect Alzheimer's Disease within Seconds

Based on these signals K.P. Subbalakshmi, the main author of the study, together with his students set out to create the aforementioned artificial intelligence tool.

For her training, the tool was nourished with data from existing transcripts from more than a thousand interviews generated as a result of an image description process to which the person suffering from Alzheimer's is subjected in order to detect the presence of abnormalities in their language associated with the disease.

Through segmented individual phrases and values assigned to them, the algorithm was able to analyze the structural and thematic relationships between them. This made the algorithm progressively able to differentiate speech from people suffering from Alzheimer's from that of healthy people, reaching an accuracy of 95%.

Post a Comment