Hydrogen is a promising energy carrier for sustainable energy systems, but its interaction with metallic structures poses significant challenges, particularly hydrogen embrittlement. Multi-principal element alloys (MPEAs), including high- and medium-entropy alloys, offer resistance to hydrogen embrittlement and potential for hydrogen storage due to their disordered atomic lattices, which create effective trapping sites for hydrogen. However, the vast compositional space of MPEAs limits experimental exploration, and conventional simulation approaches are often too computationally intensive for high-throughput screening. This thesis introduces an efficient and comprehensive computational framework for predicting hydrogen diffusivity in body-centred cubic (BCC) MPEAs. The diffusion energy landscape is characterised by statistical parameters that describe the distribution of saddle point and well-energies. Through kinetic Monte Carlo (KMC) simulations, a large dataset of hydrogen diffusivity was generated using synthetic energy landscapes defined by these statistical parameters. Machine learning symbolic regression (MLSR) was then employed to derive analytical expressions that relate the statistical descriptors to macroscopic diffusivity. To apply the model to real alloys, hydrogen diffusivity is obtained through the MLSR expressions based on energy landscape statistics that were determined using climbing-image nudged elastic band (CI-NEB) calculations with universal machine learning interatomic potentials (uMLIPs). The predictions were validated against molecular dynamics (MD) simulations, showing reasonable agreement. This framework enables fast, scalable prediction of hydrogen diffusion in complex alloys, supporting accelerated materials discovery for hydrogen-related applications.