You “sound” stressed!

LIH to collaborate with LISER on the identification of vocal biomarkers of stress

• Department of Precision Health
• Deep Digital Phenotyping Research Unit

Dr Guy Fagherazzi will participate in FragMent, a project led by Dr Camille Perchoux from the Luxembourg Institute of Socio-Economic Research (LISER), aiming to evaluate the extent to which the spatial and temporal fragmentation of exposure to environments in daily life influences physiological and psychological stress. In this context, the identification and analysis of vocal biomarkers will be used to assess the impact of urban environment, exposure and activity patterns on daily-life stress.

Stress constitutes a significant burden worldwide, with two out of three individuals reporting stress-related mental illnesses, and one out of five reporting repercussions on their physical health. Indeed, stress is a common risk factor in 75-90% of diseases, and a major contributor to mental disorders, autoimmune, infectious and cardiovascular diseases, and even some cancers. Moreover, it increases the likelihood of adopting unhealthy behaviours such as smoking, alcohol/substance abuse, physical inactivity, and unhealthy diet, further influencing chronic diseases. As the prevalence of stress increased importantly over the last decade, prevention has become a major public health priority. The FragMent project thereby aims to support policymakers in reducing exposure to environmental stressors in outdoor spaces in everyday life.

To assess the impact of urban environment, exposure patterns and activity patterns on momentary, daily and chronic stress, the study will rely on both self-reported psychological stress measurements and physiological measurements based on vocal biomarkers. In this context, the LIH Deep Digital Phenotypingteam led by Dr Fagherazzi will lend its expertise in this area to develop vocal biomarkers for stress monitoring in the general population. Study participants will be asked to perform different voice recordings, for instance while reading a pre-specified text, saying “aaaaah” as long as possible on one single breath, or counting from 1 to 20. The DDP team will subsequently process and analyse the recordings and leverage machine-learning algorithms to identify vocal biomarkers that can accurately evaluate physiological stress. These biomarkers will also be externally validated in a different cohort, namely the participants of the ongoing Colive Voice study, in order to confirm their ability to accurately detect stress.

Relying on vocal biomarkers is a non-invasive, quick and potentially less biased way to measure stress compared to questionnaires, and could be easily integrated into future digital tools for stress monitoring

explains Dr Fagherazzi, Director of the LIH Department Precision Health (DoPH) and leader of the DDP unit.

“Indeed, stress is known to influence vocal symptoms, such as a strained, tired, hoarse, low-pitched or broken voice, as well as coughing and the so-called lump in the throat. These have all been shown to be accurate indicators of physiological and psychological stress”, adds Dr Perchoux. “We are therefore glad to be collaborating with the DDP team on the identification of novel vocal biomarkers, given their established expertise in this field, as well as in machine learning”, she concludes.

The FragMent project is funded by the European Research Council (ERC) through a Starting Grant. For more information, visit the dedicated website https://www.fragmentproject.eu/