Parameter estimation of complex mathematical models of human physiology using remote simulation distributed in scientific cloud

A generic system for estimation of model parameters - calibrate models - is introduced. The proposed system architecture is built of several loosely coupled modules behaving as RESTful web services and allowing to integrate other parts of the system via HTTP protocol and data exchanged in JSON format. The system was designed in such a way that the most demanding computational part is computed in parallel and computation may be distributed to remote computational resources. A test deployment was done in scientific cloud provided by czech NGI CESNET. Parameter identification of complex models got significant speedup on cloud computing resources.