Scientists have created an AI-powered tool to predict Alzheimer's disease by analyzing speech. Alzheimer's disease, the most common type of dementia, impairs memory and thinking, affecting daily activities. Boston University researchers trained the model using audio recordings of patients with mild cognitive impairment, which is the early stage of memory loss. The model was 78.5% accurate in predicting whether patients would develop dementia within six years.
The study, published in the journal Alzheimer's and Dementia, examined initial interview recordings from 166 patients aged 63 to 97. The machine learning model found correlations between speech, demographics, diagnosis, and condition progression, focusing on spoken words and sentence structure rather than speech characteristics such as enunciation or speed.
By combining speech analysis with basic demographic information like age and gender, the model generated a score indicating whether a patient would remain stable or progress to dementia. Despite obstacles such as low-quality recordings and background noise, it performed well.
Early detection is critical, as current diagnostic tests frequently detect Alzheimer's only after significant cognitive decline. The researchers plan to make their model available through an app, potentially increasing screening availability, particularly in remote areas.
Inputs by Agencies
Image Source: Multiple Agencies
Ⓒ Copyright 2024. All Rights Reserved Powered by Vygr Media.
