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

2026-04-08

AI Summary

This study addresses the lack of objective, continuous monitoring tools for psychological stress in older oncology patients—a gap exacerbated by reliance on subjective questionnaires that are difficult to integrate into long-term cardiotoxicity management. The authors propose a novel approach combining weakly supervised multiple instance learning with heterogeneous wearable visual representations, leveraging data from smartwatches and chest-worn ECG sensors. A lightweight pretrained mixture-of-experts model (Tiny-BioMoE) extracts features, while an attention mechanism aggregates temporal information to enable end-to-end stress-level prediction without requiring per-window annotations. Evaluated on the multicenter CARDIOCARE cohort, the model achieved R² = 0.24/0.28 and Spearman’s ρ = 0.48/0.52 at months 3 and 6 of treatment, respectively, with RMSE/MAE of 6.13/5.54 at month 6, demonstrating both methodological validity and clinical promise.

Abstract

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, multicenter 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 multiple 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: R^2=0.24, Pearson r=0.42, Spearman rho=0.48; M6: R^2=0.28, Pearson r=0.49, Spearman rho=0.52), with global RMSE/MAE of 6.62/6.07 at M3 and 6.13/5.54 at M6.

Problem

• psychological stress
• elderly oncology patients
• cardiotoxicity surveillance
• wearable data
• Perceived Stress Scale

Innovation

• multi-instance learning
• wearable sensors
• visual representations
• stress estimation
• Tiny-BioMoE