Psychiatric disease diagnosis and monitoring currently suffer due to the infrequent and subjective nature of existing symptom measures, as well as the substantial heterogeneity of behavioral patterns both between individuals and within an individual over time. These issues limit not only current clinical practice, but also the ability to develop and test novel treatments. With the proliferation of consumer health trackers such as the Fitbit, there is new promise for the creation of mental health ‘vital signs’ that are robust, easy to collect, and predictive of psychiatric outcomes. Our research works to improve Obsessive-Compulsive Disorder outcomes in particular, through the development of algorithms that can detect pathological behaviors from watch movement data. This will provide cheaper and more objective metrics for OCD severity, as well as enabling real-time interventions. Towards this end, we are also working with the substantially larger UKBiobank actigraphy dataset collected from a broader population, to develop more general metrics for stereotypy of movement. Identification of stereotyped behavior within the Biobank dataset, and the population characteristics that are linked with it, will enable our own research into OCD as well as future work on other movement abnormalities.
