Can AI predict blood pressure from voice?

Scientists at Klick Labs have developed an innovative artificial intelligence (AI) technology that can predict blood pressure levels using only a person's voice. This groundbreaking technique achieved up to 84% accuracy for women and 77% for men, offering a new, non-invasive way to monitor and detect high blood pressure.

The technology leverages vocal biomarkers to provide a practical, accessible tool for early hypertension detection — potentially revolutionizing the way we approach this common yet often undiagnosed condition.

How does it work?

The AI technology developed by Klick Labs utilizes a mobile app to analyze voice recordings for signs of hypertension. In their study, 245 participants recorded their voices up to six times a day over two weeks. The app's machine learning algorithms then analyzed these recordings, focusing on subtle vocal characteristics such as pitch variability, speech energy patterns, and sound changes.

This innovative approach proved to be highly effective, with the app achieving an accuracy rate of 84% for women and 77% for men in predicting high blood pressure.

Yan Fossat, senior vice president at Klick Labs and the study's principal investigator, explained, "By leveraging various classifiers and establishing gender-based predictive models, we discovered a more accessible way to detect hypertension, which we hope will lead to earlier intervention for this widespread global health issue."

Why does it matter?

Hypertension, known as the "silent killer," affects over 25% of the global population, according to the World Health Organization (WHO). The condition is often undetected, especially in low- and middle-income countries where access to traditional blood pressure screening tools is limited. Existing methods, such as using arm cuffs or automatic devices, can be costly and require a certain level of technical expertise, posing challenges for underserved communities.

Klick Labs' voice-based technology offers a more accessible and cost-effective alternative, potentially reaching populations that traditional methods cannot. The use of voice analysis could lead to earlier detection and intervention, reducing the risk of severe complications like heart attacks, kidney problems, and dementia.

The context

This voice-based hypertension detection marks Klick Labs' expansion into broader health monitoring applications using AI. Their prior research demonstrated the efficacy of similar voice analysis techniques for detecting Type 2 diabetes, as published in Mayo Clinic Proceedings: Digital Health in October 2023. Additionally, a recent study in Scientific Reports confirmed a link between blood glucose levels and voice pitch, further supporting the potential of vocal biomarkers in healthcare.

Jaycee Kaufman, a research scientist at Klick Labs, underscored the transformative potential of this technology: "Voice technology has the potential to exponentially transform healthcare, making it more accessible and affordable, especially for large, underserved populations."

This pioneering research, detailed in the peer-reviewed journal IEEE Access, demonstrates the growing promise of using vocal biomarkers to detect hypertension, diabetes and potentially a range of other health conditions.