Upper limb impairments are often a result of neurological or traumatic injuries (e.g., stroke, traumatic brain injuries). These impairments may decrease independence in performing activities of daily living, severely affecting patients' quality of life. Growing evidence shows that increasing upper limb therapy dose for patients could improve functional outcomes and prevent their long-term deterioration. However, due to limited resources (e.g., the number of therapists) and high rehabilitation-related costs, providing a higher therapy dose to patients is challenging, both in the clinic and after discharge, and finding new models of care is therefore critical. Minimally supervised or unsupervised robot-assisted therapy (i.e., patients training with rehabilitation devices with minimal to no supervision of an external person) holds the promise of allowing an increase in therapy dose with little impact on the additional resources needed. Because of their ability to provide active support and measure a wide set of parameters, actuated (active) rehabilitation robots seem to be the best solution for this therapy setting. However, such devices are often complex to use and have never been extensively tested minimally or unsupervised. Consequently, the data available on their feasibility and cost-effectiveness is limited. At the Rehabilitation Engineering Laboratory (RELab, ETH Zurich), we developed ReHandyBot, an actuated device for unsupervised upper limb therapy. The device has already been tested with stroke patients in an unsupervised setting, with positive results regarding usability and increase in therapy dose. The main goal of this project is, therefore, to investigate the feasibility and cost of our robot-assisted continuum of care model (i.e., from inpatient to ambulatory). Specifically, we aim to evaluate the feasibility of integrating ReHandyBot into the clinical routine and perform a preliminary cost-benefit analysis of supervised, minimally supervised, and unsupervised robot-assisted therapy. Furthermore, we will also gather data on the overall dose of robot-assisted therapy and the intensity of use of the device in a clinical and ambulatory setting.