MIT Researchers: Sound of Your Voice Can Help Predict Disease

BOSTON — Your voice could help doctors diagnose everything from cancer to Alzheimer’s disease to depression.

Local researchers at MIT and Boston Children’s Hospital are joining teams from across the country in a National Institutes of Health-funded, 4-year study.

The idea is to use an app to record your voice, then researchers will compile voices nationwide and analyze them. The goal would be to find biomarkers for everything from Alzheimer’s and heart problems to depression and even cancer.

The lead researcher at MIT explained how it would work for Boston 25 News.

“It’s almost like being Sherlock Holmes to voice, taking voice as a signal and trying to understand what’s going on behind it,” said Satrajit Ghosh, a principal research scientist at MIT’s McGovern Institute. “And can we backtrack from voice and say this is ‘Disorder A’ versus ‘Disorder B’?”

Dr. Ghosh says they expect to start collecting and analyzing voice data at MIT next summer and will focus specifically here on mental health.

“It’s not anxiety in a day or depression in a day. It’s over time. So we are really hoping that the availability of smartphones and tablets in people’s homes become an avenue to collect such data that we can leverage,” Ghosh added. He says bioethicists are involved to protect patient privacy.

Similar technology has already been used to help children with mental health issues in Georgia. The Atlanta start-up TQIntelligence created an app that uses voice samples and artificial intelligence to help diagnose mental health issues in children.

The company has some big-time investors. Google, Blue Cross and the National Science Foundation have invested about $1.5 million dollars to develop this technology.

The NIH study serves as a sign of hope for families like Roger Cochran’s.

“I was terrified,” said Cochran about his Alzheimer’s disease diagnosis three years ago. “I had this vision of what it meant to have Alzheimer’s which was to be completely incompetent.”

“I saw the memory loss in his social behavior of not wanting to come with me maybe to some social functions,” said his wife, Dorothy Merrick. When talking about the study, she said, “I think that’s fantastic. I think that the more we learn about a disease, the better it can be.”

Dr. Ghosh echoed Merrick’s enthusiasm about using voice as a tool for diagnosis.

“This excitement that such a simple signal, one that we have used pretty much since the existence of humankind, becomes a rich source of understanding things that are going on inside our body,” Ghosh said.

Publication: HyunBum Kim et al., Convolutional Neural Network Classifies Pathological Voice Change in Laryngeal Cancer with High Accuracy, MDPI (2022). DOI: 10.3390/jcm9113415

Original Story Source: Massachusetts Institute of Technology