Design and Evaluation of an Open-Source Smartwatch Application for Clinical Research

Sammanfattning

Smartwatches are widely used in clinical research, yet commercial devices often rely on closed algorithms that limit transparency and data governance. Open‑source alternatives, such as the BangleJS 2, offer opportunities to develop configurable, clinical research‑oriented applications. This thesis aimed to identify researchers' requirements for an open‑source smartwatch application and to design and evaluate the application, AW app, based on these needs. Using a Design Science Research approach, we conducted semi‑structured interviews and a literature review to establish the need for a transparent, flexible, and configurable application. The identified requirements shaped the design and development of the app, which was iteratively refined through technical testing and a qualitative evaluation with the participating researcher. In total, 34 requirements were identified and prioritised using the MoSCoW model. The resulting application supports aggregated and raw data collection, sensor selection, configurable sampling frequency and RAM buffer size, EMA settings, and data streaming to web and mobile applications. Technical testing demonstrated successful local logging and stable BLE communication, while the qualitative evaluation indicated high usability. In conclusion, the AW app fulfils the core research needs identified in this study, despite some remaining limitations, and provides a usable foundation for future clinical research. Further work may extend the EMA functionality and implement failure detection to support more advanced study designs, to take it from a proof-of-concept to a more robust and usable application.