Data-driven Digital Therapeutics Analytics

Digital therapeutics (DTx), in contrast to traditional treatments such as pills, use software installed in smartphones or wearable devices as a medical device to cure diseases and improve health conditions, which represents a significant departure from existing wellness products such as Fitbits. DTx requires clinical validation of efficacy through systematic clinical trials, as do conventional therapeutics. Mobile DTx apps transform conventional treatment approaches such as counseling, self-help, and self-tracking into app-based micro-interventions that can be delivered via notifications, short videos, and chatbots. This article presents a data-driven DTx analytics framework for analyzing and optimizing DTx delivery processes in everyday life contexts by leveraging passive sensor data analysis and human-in-the-loop interaction support.

I. Motivation

In traditional drug delivery systems, it was easy to evaluate the effects of drugs in controlled laboratory environments. However, in the case of mobile DTx apps, it becomes challenging to evaluate and improve the efficacy of DTx, because patients use mobile apps to receive micro-interventions in an uncontrolled daily environment [1]. Since typical clinical trials only look at differences in endpoints (e.g., patients’ weight for weight management), it is non-trivial to examine which intervention components of the mobile DTx app were used and how they made the differences in the endpoints. Not surprisingly, such information in practice is essential for DTx improvement.