October 17, 2024

Excuse Me, What Did You Say?

Monitoring fluctuations in emotion, mood, and functioning through changes in speech.

We’re working on making it easier to tell how severe someone’s mood symptoms are by listening to how they talk, especially for people with bipolar disorder (BD). BD involves shifts in mood from normal to either really high (mania) or really low (depression), which can significantly affect someone’s life. Our goal is to find ways to spot these changes in mood as they happen, without making people do anything extra or different. We’re inspired by the idea that you can often detect changes in someone’s mood through the way they sound, so we’re developing technology that listens to people’s voices and picks up on emotional changes.

Thanks to funding from the National Institutes of Health (NIH), we’re planning to study how 160 people talk in their daily lives over the next five years using smartphones equipped with an app called PRIORI. The participants will also tell us how they’re feeling and we’ll have clinical information about their mental health. We’re using advanced AI to analyze the emotions in their voices. We’re trying to figure out how these emotional clues, along with the participants’ own reports of their feelings, can help us understand the severity of their mood swings, whether they’re feeling manic or depressed.

The study we’re doing with the NIH is built on what we learned from an earlier, smaller, research project, which was funded through generous philanthropic support from Patricia and Jerry Wagner, Jan and David Baszucki, and The Richard Tam Foundation. In that study, we used the PRIORI app to collect voice data and discovered that we could predict how severe a person’s mood symptoms might be. For instance, our predictions matched very closely with how the participants said they were feeling each day in terms of depression. Additionally, the information we got from analyzing their speech and the emotions they reported feeling contributed to our prediction model in similar ways. The long-term goal of this project is to provide innovative AI tools that people with bipolar disorder can use to help manage their illness and improve outcomes.