Cancer patients experience substantial functional impairments during and after treatments, often requiring rehabilitation and exercise interventions. However, a functional assessment of the specific deficits and impairments in the individual is needed to inform exercise therapy. The oncology workforce is challenged in its capacity to conduct such functional assessment, therefore limiting the ability of rehabilitation providers to identify and meet the needs of these individuals. This work seeks to develop a novel, low-cost PC/camera-based image/video-monitoring system that can a) continuously and unobtrusively monitor the physical health of a cancer patient in his/her natural environment, b) determine the existence and extent of functional deficits, and c) facilitate personalized rehabilitative interventions, providing a risk-stratified and personalized approach to cancer patient care. Based on patient’s daily activities or exercises, objective and quantitative clinically-relevant and laboratory-validated measures of physical performance (e.g., gait speed, balance, and fall risk) will be extracted by using state-of-the-art image-processing and machine-learning techniques, involving human-detection, pose-estimation, and deep-learning algorithms and fall-risk prediction models. By providing the patients and their clinicians with a real-time assessment of their physical performance, through a PCbased software and a Web portal, respectively, the system will facilitate proactive rehabilitation via clinician-approved, personalized health-care tips and exercise recommendations.