PhD student • University of Rhode Island • ASSET Lab
I research smartphone-based digital phenotyping, using passive smartphone sensors to model health-related behaviors and outcomes. I develop applied machine learning systems that emphasize interpretability and methodological rigor to improve health outcomes for marginalized populations.
I am a fourth-year Computer Science PhD student at the University of Rhode Island and a member of the ASSET Lab led by Krishna Venkatasubramanian. My research focuses on smartphone-based digital phenotyping, which uses passive smartphone and wearable sensor data to model health-related behaviors and outcomes.
I develop applied machine learning systems that leverage multimodal sensing streams, including accelerometer, GPS, communication logs, and wearable physiological signals, into clinically meaningful behavioral representations. My work emphasizes interpretability and methodological rigor to ensure that models not only perform well but also provide transparent and actionable insights that improve health outcomes.
Developing SmartSteps, a translatable model that adds context-awareness to adherence data and anticipates potential PrEP nonadherence to deliver corrective feedback before nonadherence events occur.
Led by Dr. Peter R. Chai (Brigham and Women’s Hospital & Harvard Medical School) in collaboration with Brigham and Women’s Hospital and Fenway Health. Funded by an NIH subaward via Brigham and Women’s Hospital.
Developing wearable physiological sensing methods to detect medication non-adherence during medication-assisted treatment for opioid use disorder, reducing relapse risk by identifying non-adherence early.
Funded by an NIH NIBIB R01 grant.
Selected publications (authored/co-authored by Arlen Dumas).
Email is best: ardumas@uri.edu