Maintaining good glycemic control is a continuous challenge for type-1 diabetic patients. The current means of insulin therapy are seen to subject patients to hyper- and hypoglycemic episodes. With the advancement of computer science and technology, an artificial pancreas, a computerized device that will automatically control patient's blood glucose level by providing the substitute for the insulin supply functionality of a healthy pancreas was proposed. The main challenge faced by the development of this device is a fully closed-loop control system. In this paper we proposed an artificial pancreas system with an adaptive closed-loop control strategy that computes the appropriate insulin infusion rate and thereby keeping the patient's blood glucose concentration within normoglycemic range. The adaptability is achieved through a rigorous pattern recognition technique with patient-specific glucose readings obtained through glucose monitor at a given sampling step. The performance of the control strategy is assessed using the simulation results carried out under various physiological disturbances and meal intake. © 2012 IEEE.