Figure 3.

Schematic workflow of Roadmap 2.0. Firstly, (a) large volumes of wearable sensor (i.e., Fitbit) data stream (e.g., heart rate, sleep, activity/steps), Electronic health Records and physiological data streams will be captured in real-time in mHealth platform Roadmap 2.0. The captured multi-parameter data streams (b) will be stored in secure HIPPA compliant server. It will contain multivariate physiological signals and patient reported outcomes data (generated from patients’ response of survey questionnaires). (c) The stored data will be processed in data analytics pipeline. Here, firstly features will be extracted from all diverse types of data and then machine learning algorithms will be used to build a predictive model. This model will be applied to test set for predictions on the unseen data. Finally, the predictive model will be evaluated using AUC. Also, feature importance will be computed. (d) The final results will be stored in the secure server.