John Torous1,2, Ian Barnett3, Patrick Staples3, Luis Sandoval1, JP Onella3, Matcheri Keshavan1,2; 1Beth Israel Deaconess Medical Center, 2Harvard Medical School, 3Harvard School of Public Health
The SMART study is a pilot investigation into digital phenotyping for schizophrenia and related psychotic disorder. Utilizing the Beiwe platform that runs on patients' personal smartphones, the app can collect active data in the form of surveys and passive data in the form of real time GPS, accelerometer, anonymized call/text logs, screen use time, power status, and other features. The goal of the SMART study was to 1) assess feasibility of this method of data collection in participants with schizophrenia and 2) collect pilot data regarding the utility of digital phenotyping to aid in relapse prediction in schizophrenia Patients in active treatment and with a diagnosis of schizophrenia or a related psychotic disorder were recruited and asked to use the Beiwe app for up to 3 months. We used anomaly detection methods to calculate relapse risk based on the smartphone passive data as well as survey responses. 17 research participants used the smartphone app for up to 3 months. Three research participants experienced a clinical relapse that was recorded by the app. No participants experienced any adverse event or negative outcomes during the course of the study related to app use. While these pilot results preclude reporting on the sensitivity and specificity of digital phenotyping in relapse prediction, they do suggest this method is acceptable to patients with schizophrenia and related disorders. Our group is currently conducting a larger scale study to better characterize this method of relapse prediction.
