Abstract

Social, behavioral, and cognitive phenotypes have traditionally presented difficulties for the health sciences both in research and clinical settings. The standard approach has been to rely on self-administered surveys and clinician-administered tests, both of which tend to be subjective. Due to the burden these assessments impose on participants, this approach is usually only applicable cross-sectionally, which limits its utility, especially in areas that are characterized by episodic changes in behavior. Digital phenotyping attempts to harness continuously collected raw data from personal digital devices, especially smartphones and wearables, to address this so-called phenotyping problem. I will talk about some key concepts in digital phenotyping, provide a review of common datatypes, and illustrate some of the various methodological challenges. I will conclude by providing some examples of collaborative research in neurological diseases.