The CARP Mobile Sensing Framework -- A Cross-platform, Reactive, Programming Framework and Runtime Environment for Digital Phenotyping

Mobile sensing - i.e., the ability to unobtrusively collect sensor data from built-in phone sensors - has long been a core research topic in Ubicomp. A number of technological platforms for mobile sensing have been presented over the years and a lot of knowledge on how to facilitate mobile sensing has been accumulated. This paper presents the CARP Mobile Sensing (CAMS) framework, which is a modern cross-platform (Android / iOS) software architecture providing a reactive and unified programming model that emphasizes extensibility, maintainability, and adaptability. Moreover, the CAMS framework supports sensing from wearable devices such as an electrocardiography (ECG) monitor, and configuring data transformers. The latter allows to transform collected data to a standardized data format and to implement privacy-preserving data transformations. The paper presents the design, architecture, implementation, and evaluation of CAMS, and shows how the framework has been used in two real-world mobile sensing and mobile health (mHealth) applications. We conclude that CAMS provides a novel cross-platform application programming framework which has proved mature, stable, scalable, and flexible in the design of digital phenotyping and mHealth applications