We propose a whole person-centered approach for the development of artificial pancreas devices (AP) that automates insulin delivery for adolescents and young adults with type-1 diabetes. The proposed approach will enhance existing AP devices by means of "smart nudges" based on real-time indicators of planned and ongoing activity, cognitive load, and psychosocial measures like mood and stress. These nudges will help individuals with type 1 diabetes adapt their behaviors such as meals, physical activity and insulin bolusing to the AP device in order to maintain their blood glucose levels inside a tight "euglycemic" range while avoiding adverse events linked to extremely low and high blood glucose levels. Our proposed person- centered artificial pancreas (PCAP) approach will enhance existing control systems to reflect more nuanced understandings of users’ physiological, cognitive, psychosocial, and behavioral states and support users’ lived experiences of managing chronic conditions in continuous collaboration with assistive devices. Using real-time data on the physiological state (blood glucose, heart-rate, physical activity, and illness); behavioral data from user interactions with the device; and measures of cognitive load, stress, attention and trust obtained through carefully designed short questionnaires, PCAP will build whole person models that track mental states including situational awareness, cognitive load, attention, and stress in order to predict future behaviors. These models will be used by a decision-making algorithm to determine the parameters for a nudge, including content, importance and frequency. Our multidiscplinary team will also investigate the design of a user-interface for delivering these nudges and tracking the user response to them. A series of feasibility/preclinical user studies involving adolescents are proposed in order to evaluate the correctness, reliability, and efficacy of the proposed PCAP system. Important longer-term issues surrounding trust and privacy will be carefully investigated to inform the design of PCAP. The proposed multicomponent cognitive models will incorporate ideas from a variety of fields including human–computer interaction, psychology, mobile systems, probabilistic modeling, inference, learning and control, with a particular focus on establishing an empirical basis for effective patient "nudging" to improve diabetes self- management without increasing workload or drawing undue attention to the patient’s condition. While the focus of the project is on the treatment of type-1 diabetes, the proposed fundamental techniques will extend to the management of other chronic conditions where the integration of wearable sensors and mobile devices as part of multicomponent interventions can also guide the adoption and maintenance of healthy behaviors. The proposed research will also investigate important aspects of user privacy and ethical considerations in assistive medical devices like PCAP, given the...