New Study Demonstrates AI-Based Voice Biomarkers Can Revolutionize Depression Screening in Primary Care

Berkeley, CA — January 13, 2025 — Researchers from Kintsugi Mindful Wellness, Inc., the University of California, Berkeley, and the University of Arkansas for Medical Sciences (UAMS) have published a groundbreaking study in the January/February 2025 issue of the Annals of Family Medicine. The study, “Evaluation of an AI-Based Voice Biomarker Tool to Detect Signals Consistent With Moderate to Severe Depression,” highlights the potential of artificial intelligence (AI) to enhance mental health diagnostics through the innovative use of voice biomarkers.

Conducted between February 2021 and July 2022, the study evaluated the performance of Kintsugi’s AI technology in detecting depression by analyzing the short recordings of free-form speech. Voice samples (~25 seconds in length) were collected from over 14,000 participants across the U.S. and Canada. Kintsugi’smodel achieved a sensitivity of 71.3% and specificity of 73.5% in identifying individuals with moderate to severe depression, benchmarked against PHQ-9 scores. 

‍Given that depression screening rates currently fall below 4% in primary care, Kintsugi’s AI tool offers a non-invasive, scalable, and reproducible method that bridges gaps in mental health care across diverse clinical and virtual settings. 

“Depression is one of the most significant global health challenges, yet primary care screenings remain vastly underutilized,” said Dr. Prentice Tom, lead clinician and contributor from Kintsugi Mindful Wellness. “This research underscores the transformative potential of AI in expanding access to mental health care and improving diagnostic accuracy.”

‍The study demonstrates that machine learning tools can augment traditional methods, empowering clinicians with additional resources to enhance patient care. Researchers also noted areas for future exploration, including tailoring tools to specific demographic groups and integrating these systems into primary care workflows.

About the Research

This collaborative effort combines cutting-edge AI research from Kintsugi with academic expertise from UC Berkeley and UAMS. Supported by diverse datasets and rigorous validation protocols, this study represents one of the largest-scale applications of machine learning in mental health diagnostics to date.

The study is available online on the Annals of Family Medicine website. Read the full study here.

Media Enquiries

Craig Corbett, PR representative for Kintsugi Mindful Wellness: craig@publicize.co 

About Kintsugi

Kintsugi is a leader in voice biomarker AI, enabling early detection of clinical depression and anxiety from just 20 seconds of speech. Its API platform integrates seamlessly into call centers, telehealth visits, and remote patient monitoring applications, empowering clinicians to deliver timely, evidence-based mental health care while improving outcomes for patients. Headquartered in Berkeley, California, Kintsugi is dedicated to making mental health care more accessible and effective for individuals worldwide.

‍

‍

‍