The past few years have witnessed a rapid advancement of mobile healthcare systems. However, in the mobile computing environment, the resource fluctuations, stringent application requirements and user mobility have severely hindered the performance and reliability of the healthcare service delivery. The current approaches to solve this resource supply and service demand mismatch problem either limit the adaptation to an isolated node or require significant user's involvement. Given the distributed processing paradigm of mhealth system, we propose that a new adaptation approach could be dynamically redistributing processing tasks across distributed nodes. This PhD research addresses two main issues to validate this approach: (1) computation of a suitable assignment of tasks at compile-time or run-time; and (2) dynamic distribution of tasks across the nodes according to this new assignment at run-time.