Precise sorting of target cells is vital for the progression of cellular research and the enhancement of medical diagnostics. Inertial microfluidics is a recent development in technology providing a very promising approach for high throughput, label-free particle separation. This study presents an extensive CFD numerical investigation on particle migration and segregation within a symmetric serpentine microchannel. The research applies the Eulerian framework to resolve fluid dynamics and the Lagrangian approach to trace particle trajectories, investigating the influence of flow Reynolds number and serpentine channel loops on sorting efficacy. A numerical set of CFD simulations that were executed over 200 randomly set-up data configurations was used for constructing a robust data-driven model. Based on the dataset, a machine learning model was developed to forecast inflow parameters representative of specific sorting efficiencies. In particular, 92% prediction accuracy for the Channel Reynolds Number was achieved in validation. Nevertheless, predicting the number of loops that will ensure optimal sorting performance is still quite a challenge.