A multiple sclerosis two-compartmental differential equation computational model 3D simulation using OpenCL

Abstract

Expanding on our previous conference paper (de Paula et al., 2023), this work introduces a novel two-compartmental 3D mathematical model based on differential equations to simulate Multiple Sclerosis (MS) dynamics. The mathematical model incorporates key MS processes like lymphocyte infiltration, antigen presentation, adaptive immune response activation, and demyelination. Implementing such a multi-scale, 3D problem is inherently complex. To address this, we utilised a heterogeneous computing environment combining CPUs and GPUs. However, this environment introduces load-balancing challenges. Initially, we tackled these challenges by employing two distinct load-balancing approaches to optimise simulation performance. Results reveal performance improvements of up to $4.4\times$ compared to the non-load-balanced version.