How spontaneous speech changes due to Alzheimer’s disease

A major goal of our research is to try to better understand how spontaneous speech—everyday conversation—may provide clues that a brain is beginning to experience changes due to Alzheimer’s disease.  

Modern clinical care is often rushed, with little time for a thorough evaluation, much less the kind of detailed clinical examination that most people recommend for a diagnosis of Alzheimer’s.  Yet many different people now from all over the world are demonstrating how computers can distinguish those with dementia from those without.  Accuracy of these programs approaches 90%– similar to human physicians when it comes to recognizing Alzheimer’s disease. However, there’s a number of steps that would be necessary before this technology can be clinically useful, and the technology should only be used only in certain, protected circumstances, such as a doctor’s visit. 

For example, most studies only look at extremes of dementia, rather than complex or subtle cases that are more common day to day.  It will also help to understand why the computer has identified certain cases as dementia—algorithms that do this are not always easily understood by doctors or patients, who are unlikely to just place blind faith in the machine’s decision. In this lab, we’re working to understand why a computer algorithm may decide someone has dementia based on their speech.