How AI Can Detect Children with Anxiety and Depression
Researchers create an AI voice screening tool for children.
Posted May 08, 2019
Recently University of Vermont researchers and their colleagues announced the creation of an artificial intelligence (AI) screening tool that can detect both anxiety and depression in children from their voices with a high level of accuracy. They published their results in April 2019 in IEEE Journal of Biomedical and Health Informatics.
Children, like adults, experience anxiety and depression—internalization disorders. However, unlike adults, young children often have difficulty communicating their symptoms, making it challenging for health care professionals to accurately diagnose and treat. Psychiatric symptoms that begin early in life run the risk of becoming chronic issues through adolescence and adulthood. Having a tool that can objectively screen children for anxiety and depression is a valuable asset for early childhood intervention in order to prevent future problems that can lead to substance abuse, suicide risk, and other mental health issues.
Nearly 20 percent of children experience an internalization disorder of anxiety or depression, according to the research team of Ellen W. McGinnis, Maria Muzik, Kate Fitzgerald. Katherine L. Rosenblum, Jessica Hruschak, Ryan McGinnis ,Steven P. Anderau, Reed D. Gurchiek, and Nestor L. Lopez-Duran.
Using a modified version of the Trier-Social Stress Task, an assessment tool that induces stress and anxiety in the test taker, the researchers recorded the audio of 71 children between the ages of three and eight who were tasked with creating a three-minute story that would be judged on interest. A buzzer would sound after 90 seconds and again when there was 30 seconds remaining. The children were also evaluated using standard methods—a clinical interview and parent questionnaire.
The audio recordings were fed into an AI machine learning algorithm to analyze the statistical features. The team discovered that three audio features in particular were highly indicative of identifying internalization disorders—low-pitched voices, higher-pitched buzzer responses, and repeatable speech inflections and content.
The researchers wrote, “We show that machine learning analysis of audio data from the task can be used to identify children with an internalizing disorder with 80% accuracy (54% sensitivity, 93% specificity),” and that “the algorithm requires just a few seconds of processing time once the task is complete to provide a diagnosis.”
The researchers report results that “point toward the future use of this approach for screening children for internalizing disorders so that interventions can be deployed when they have the highest chance for long-term success.”
As next steps, the team plans to further develop the speech analysis algorithm for clinical use to help detect anxiety and depression in children based on voice analysis. The hope is that their work will pave the way in one day providing an objective and rapid method of diagnosing mental health conditions that are not only difficult to identify, but also often overlooked in children.
Copyright © 2019 Cami Rosso All rights reserved.
McGinnis, E. W., et al., "Giving Voice to Vulnerable Children: Machine Learning Analysis of Speech Detects Anxiety and Depression in Early Childhood." IEEE Journal of Biomedical and Health Informatics. April 26, 2019.