A group of researchers from the Luxembourg Institute of Health and the University of South Florida has established the first consensus-based framework to standardise vocal biomarkers for use in artificial intelligence-based disease detection.
The study, which was published in the journal Digital Biomarkers as part of the VOCAL initiative, aims to resolve the lack of consistent terminology currently used in voice-based medical research. Previously, terms such as voice, speech, and vocal biomarkers were often used interchangeably, complicating the work of clinicians and regulators.
The new framework introduces a hierarchical model that covers the various domains involved in voice and speech production. It specifically distinguishes between general vocal measures and validated vocal biomarkers to provide a more precise scientific language.
The collaborative effort involved 24 experts from Europe and North America. The goal was to foster better coordination between a diverse range of stakeholders, including data scientists, engineers, speech specialists, and industry regulators.
This standardised approach is intended to accelerate the development of digital health technologies for diagnosing and monitoring several medical conditions. These include Parkinson’s disease, Alzheimer’s disease, depression, heart failure, and type 2 diabetes.
