Engineering Large Wearable Sensor Data towards Digital Measures

Developing digital biomarkers requires handling unprecedented quantities of digital data generated from digital health technologies that utilize a combination of computing platforms, connectivity, software, and sensors. These collected data need to be transformed and transported into a meaningful and useful format before further being derived into health indicators for understanding disease state and life quality. The unique challenges for this class of data engineering tasks are due to the complexity and volume of the digital data we are handling, the data quality and fidelity required to enable subsequent analysis, and the repeated cycles of trial and error to achieve the desired results. This paper presents a family of systems, pipelines, and methods we have designed and built to facilitate these tasks in a typical digital data engineering lifecycle in the context of digital biomarker development.