OPEN: A Benchmark Dataset and Baseline for Older Adult Patient Engagement Recognition in Virtual Rehabilitation Learning Environments

Engagement in virtual rehabilitation programs is crucial for participant satisfaction, adherence, and successful program delivery. Measuring engagement in virtual group settings remains a significant challenge. There is growing interest in applying artificial intelligence (AI) for large-scale, automated engagement recognition. While engagement has been extensively studied among young learners in academic contexts, research methods and datasets focusing on older adults in virtual and telehealth environments are limited. Many existing approaches overlook the contextual relevance of learning materials and the longitudinal nature of engagement. This paper introduces OPEN (Older adult Patient ENgagement), a novel dataset created to support the development of AI models for engagement recognition. Data were collected from eleven older adults over six weeks of virtual cardiac rehabilitation sessions, totaling nearly 35 hours, making it one of the largest datasets of its kind. To protect privacy, raw video is withheld, but released data include facial, hand, and body joint landmarks, along with behavioral and affective features. Annotations provide second-level binary engagement states, emotional and behavioral labels, and contextual details. Machine-learning and deep-learning models trained on the dataset achieved up to 81% accuracy. OPEN provides a scalable foundation for advancing AI-driven engagement recognition in aging populations and broader research contexts.