Fast Parameter Estimation for Cancer Cell Progression and Response to Therapy

We investigate a mathematical model describing the development of cancer cells in the mammalian cell division cycle under therapy and present an efficient strategy for fast numerical simulations and effective treatment programs in parallel computing environments. We perform a series of computations implemented on a cluster of computers with multicore processors for which the model equations are algorithmically split and solved over independent processors working in parallel and examine the speedup gained. In our implementation, the time interval is split into a number of subintervals corresponding to the number of available processors and the parallelization is invoked across time so that the number of processors to be used is unrestricted. We also extend our implementation to a generalized model of human tumor growth in vivo and demonstrate the computational efficiency of the algorithm and the associated decrease in computational time as the number of utilized processors increases by means of a series of numerical experiments performed in a parallel computing environment.