App Could Detect Alzheimer’s Disease Through Phone Calls
Researchers have recently applied machine learning to create models that use acoustic aspects of a person’s conversations to determine whether they have early Alzheimer’s—a new research shows.
Alzheimer’s disease is characterized by progressive degradation of the brain areas that control memory, thoughts, and language. According to the CDC, up to 5.8 million people in the US lived with the disease in 2020.
Early diagnosis, according to research, is vital since it allows clinicians to begin clinical interventions soon to manage the person’s symptoms. However, there are currently no affordable, widely accessible, and reliable techniques for diagnosing Alzheimer’s disease in its early stages.
One possible diagnostic indicator is that patients with the condition talk more slowly in everyday conversation, stopping as they strive to find the right words. This can result in speech that may be less fluent than that of those who don’t have the disorder.
Researchers from Tokyo Medical and Dental University, McCann Healthcare Worldwide, Kyoto University, and Keio University believe that a fully automated model could predict who is likely to develop Alzheimer’s disease by analyzing the acoustic features of speech, such as pauses, pitch, and voice intensity.
Using machine learning, scientists created models they believe will someday be as good as, if not better than, a routine test used by clinicians to diagnose the disease.
The researchers analyzed voice data from 24 individuals with Alzheimer’s disease and 99 people without the condition, all of whom were 65 years old or older, using three machine-learning algorithms.
The audio recordings were taken from a Hachioji public health initiative in which participants talked on the phone about making lifestyle changes to lower their dementia risk.
In conclusion, the researchers suggest that developers implement their model into websites or mobile apps, making it available to the entire public. They believe that such a tool could aid people seeking professional care in the early stages of the condition.
However, researchers would need to test the model on a larger sample of the general population and then track them over time to see who developed the condition to prove that it works.
Image by Sabine van Erp from Pixabay