A new AI-based model can predict the onset of Alzheimer’s disease by analysing a person’s speech, developers said.
According to the researchers, a model trained on audio recordings of patients with mild cognitive impairment (the early stages of memory loss) achieved 78.5 percent accuracy in predicting whether the patients would remain stable or progress to dementia within six years.
Alzheimer’s disease is the most common form of dementia and affects a person’s everyday activities by impairing memory and thinking.
Researchers at Boston University in the US used recordings of initial interviews of 166 patients aged 63-97 years and trained a model using machine learning to understand patterns between speech, demographics, diagnosis and deterioration of their condition.
The study, published in the journal Alzheimer’s & Dementia, showed that the model analysed interview content, such as words spoken and sentence structure, but not speech characteristics such as accent or speed.
“We combine the information from the audio recording with some very basic demographics — age, gender, etc. — and we get a final score,” said Ioannis C. Paschalidis, professor of engineering and corresponding author of the study.
“You can look at this score as the probability that someone will remain stable or progress to dementia. It had significant predictive ability,” Pascalidis said.
The model was able to perform well despite challenges such as low-quality recordings and background noise, the researchers said.
The researchers emphasized that early prediction is very important, as current diagnostic tests often can only identify Alzheimer’s disease when significant decline in cognitive ability has already occurred, such as memory loss and changes in personality traits.
The team aims to make its model accessible to patients living in remote areas through an app, which will increase the number of people getting tested.
(Except for the headline, this story has not been edited by NDTV staff and is published from a syndicated feed.)