Manolis Tsiknakis

Early prediction of therapy-induced cardiotoxicity from a single echocardiography video remains a significant challenge, despite strong performance in other cardiac assessments.

TDGT achieves high-fidelity synthetic data generation with minimal manual tuning, revolutionizing how we approach privacy-preserving data sharing in AI.

Wearable data, particularly smartwatch activity and sleep features, can predict frailty-related functional changes in elderly cancer patients with surprisingly high accuracy, offering a more continuous and objective assessment than infrequent clinic visits.

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.