Remote assessment and monitoring of depression
Many efforts have been made to identify markers representative of treatment response, but treating mental health remains a trial-and-error approach. The proliferation of mobile devices and wearable sensors for behavioral and physiological monitoring could help address these challenges. The use of digital sensors can enable continuous, unobtrusive, and objective assessment of disease severity and progression, allowing treatment personalization through the identification of relevant digital biomarkers.
Involved People
Digital phenotyping, first defined in 2016 as the “moment-by-moment quantification of the individual-level human phenotype in situ using data from personal digital devices”, exploits these data to develop “digital biomarkers” for explaining, influencing, and/or predicting health-related outcomes. While we are still in the early days of digital phenotyping, it has seen a growing interest in psychiatry, given the widespread of smartphones and wearables and their potential to create data-driven, objective measurements of behavior, mood, and mental states.
A great body of influential work suggests that passive and mobile sensing data, such as physiological and behavioral signals, can be integrated with Machine Learning (ML) to diagnose mental health disorders. ML has the potential to offer data-driven assessments from continuous and non-invasive collected data, which are more objective, accurate and reliable, and has the potential of increasing efficiency while reducing costs in the field of mental health. Mental health treatments and interventions typically follow a one-size-fits-all approach, looking at those treatments that explain the benefits and variance for the “majority of a clinical group” and test for “group effects”. Multi-modal data combined with machine learning can lead to precision medicine by providing differential diagnosis and individualized treatment.
In this study, we will remotely collect continuous behavioral and physiological data from depressed patients during their daily life for several weeks. Through Machine Learning models we will extract relevant features from this longitudinal multimodal dataset for the design of digital biomarkers of depression symptoms.
Funding
SNF Ambizione, “Digital Tools for Mental Health: Closing The Loop for Personalized Treatment” (#193291)