Stress Estimation in Elderly Oncology Patients Using Visual Wearable Representations and Multi-Instance Learning

Psychological stress is clinically relevant in cardio-oncology, yet it is typically assessed only through patient-reported outcome measures (PROMs) and is rarely integrated into continuous cardiotoxicity surveillance. We estimate perceived stress in an elderly, multi-center breast cancer cohort (CARDIOCARE) using multimodal wearable data from a smartwatch (physical activity and sleep) and a chest-worn ECG sensor. Wearable streams are transformed into heterogeneous visual representations, yielding a weakly supervised setting in which a single Perceived Stress Scale (PSS) score corresponds to many unlabeled windows. A lightweight pretrained mixture-of-experts backbone (Tiny-BioMoE) embeds each representation into 192-dimensional vectors, which are aggregated via attention-based multi-instance learning (MIL) to predict PSS at month 3 (M3) and month 6 (M6). Under leave-one-subject-out (LOSO) evaluation, predictions showed moderate agreement with questionnaire scores (M3: R2=0.24, Pearson r=0.42, Spearman ρ=0.48; M6: R2=0.28, Pearson r=0.49, Spearman ρ=0.52), with global RMSE/MAE of 6.62/6.07 at M3 and 6.13/5.54 at M6. Clinical Relevance— This work provides a scalable pathway to estimate patient-reported stress from passively collected wearable data in elderly cardio-oncology, enabling longitudinal monitoring between clinic visits. Even moderate predictive skill can support holistic surveillance by contextualizing stress alongside sleep disruption, activity decline, and cardiac dynamics during cardiotoxic treatment cycles. 

Index Terms—Multiple Instance Learning, Mixture of Experts, Stress estimation, Wearable Sensing, Cardio-oncology, Multimodal
Learning, Visual Representations