The proposed pilot study will use passive data collected by mobile devices (i.e., smartphones and wearable direct-to-consumer fitness trackers with built-in heart rate monitors) to predict ecological momentary assessment of psychological stress during cancer treatment. The goals of the proposed research are (1) to test the feasibility and acceptability of using smartphone and wearable wrist sensors to predict stress ratings during chemotherapy for colorectal cancer and (2) to develop preliminary machine learning models to predict patient-reported stress from passively sensed data. The proposed study represents the first effort to integrate passive smartphone and physiological sensor data to predict stress as well as an initial attempt to develop automated stress detection models in the context of active cancer treatment, a time of significant and fluctuating daily psychological stress. The pilot data collected from 25 patients over four weeks of chemotherapy will be critical to establish feasibility and proof-of-concept for this work. If feasibility and preliminary models are supported, these data will guide an external funding application to develop and test a real-time behavioral intervention to reduce stress during cancer treatment.