Assessing mental disorders with digital biomarkers

Mental disorders such as depression and anxiety significantly contribute to the global disease burden. The World Health Organization estimates that mental disorders affect one in eight people globally. Mental disorders lead to adverse health outcomes and have a direct cost impact on society. Despite the availability of effective therapy and medication, a key challenge in diagnosing, monitoring and treating mental disorders is the inadequacy of assessment methods.

This article-based doctoral thesis develops and investigates the feasibility of tools leveraging smartphones and wearables, statistics and machine learning technology to augment the traditional methods of mental disorder care. We developed a smartphone-based application for passive and active data collection leveraging embedded smartphone sensors and a data analysis and behaviour modelling pipeline for quantifying digital biomarkers from smartphone data, predictive analysis, and monitoring mental disorder symptoms.

We found statistically significant differences in digital biomarkers and moods of people with and without symptoms of depression. We found a statistically significant relationship between digital biomarkers, mood, and symptoms of depression and anxiety. We show that digital biomarkers and mood can predict symptoms of depression, and that it is feasible to passively monitor fluctuations in mental disorder symptoms to inform clinical decisions.

The key findings in this thesis show the feasibility of augmenting the current mental disorder care methods with evidence-based and continuous assessment of symptoms in general and clinical populations in everyday life. The tools developed in this thesis could be tailored for various mental disorders such as schizophrenia, post-traumatic stress disorder, bipolar disorder, and anomalous human behaviours such as sleep disorders, sedentary behaviours and problematic smartphone use. Collaborating with public health policymakers and clinicians, we see the potential to impact mental disorder care with just-in-time clinical interventions based on automated early detection of mental disorders and flagging deterioration of mental disorder symptoms.