Search papers, labs, and topics across Lattice.
This paper explores estimating perceived stress in elderly breast cancer patients using wearable sensor data (ECG, activity, sleep) transformed into visual representations. A pre-trained mixture-of-experts model (Tiny-BioMoE) embeds these representations, and attention-based multiple instance learning (MIL) aggregates the embeddings to predict Perceived Stress Scale (PSS) scores. The method achieves moderate agreement with questionnaire scores (R^2 up to 0.28) in a leave-one-subject-out evaluation, demonstrating the potential of wearable data for continuous stress monitoring in oncology.
Wearable sensor data, transformed into visual representations and processed with attention-based MIL, can predict perceived stress in elderly cancer patients with moderate accuracy, offering a less intrusive alternative to questionnaires.
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.
