Improved upper limb rehabilitation requires careful and re-constructed information around stroke patients' muscle activation characteristics and kinematic features in functional movement. Body Sensor Networks (BSN) are deployed to provide an immersive engagement of the rehabilitation exercise and translation into an augmented reality world for a higher order of analytics and consultation by medical consultants. Results of the analysis generate contextual intelligence to improve therapy programmes in order of an increased magnitude with derived information on model schemas, pattern deviation and effectiveness of diagnostics.