Measuring Stress in Health Professionals over the Phone using Automatic Speech Analysis during COVID-19 Pandemic : Observational Study (Preprint)
BACKGROUND During the current COVID-19 pandemic, health professionals are directly confronted with the suffering of patients and their families. By making them main actors in the management of this health crisis, they are exposed to various psychosocial risks (stress, trauma, fatigue, etc.). Paradoxically, stress-related symptoms are often underreported in this vulnerable population but potentially detectable through passive monitoring of changes in speech behavior. OBJECTIVE The study aims to investigate the use of a rapid and remote measure of stress levels in health professionals working during this COVID 19 outbreak through the analysis of their speech behavior during a short phone call conversation, and in particular a positive/negative and neutral story telling task. METHODS For this, speech samples of 89 healthcare professionals were collected over the phone and various voice features extracted and compared with classical stress measures via standard questionnaires. Regression analysis was additionally performed. RESULTS Certain speech characteristics correlated with stress levels in both genders; mainly spectral (formant) features as the Mel-frequency cepstral coefficients (MFCC) and prosodic characteristics such as the fundamental frequency (F0) seemed sensitive to stress. Overall, for both male and female participants, using vocal features from the positive tasks for regression yielded most accurate prediction results of stress scores (MAE = 5.31). CONCLUSIONS Automatic speech analysis could help with early detection of subtle signs of stress in vulnerable populations over the phone. Combining the use of this technology with timely intervention strategies it could contribute to the prevention of burn outs as well as the development of co-morbidities such as depression or anxiety.