This article explores advanced optimization strategies for XCMS parameter tuning, focusing on metaheuristic methods such as Genetic Algorithms (GA) and Bayesian Optimization (BO). Unlike IPO, which relies on isotopologue patterns, these methods navigate complex parameter spaces through iterative search and probabilistic modeling. We compare their strengths and limitations, highlighting trade-offs between exploration capability and computational efficiency. Finally, we discuss emerging deep learning–based approaches that aim to predict optimal parameters directly, suggesting a shift from iterative search toward data-driven automation in metabolomics workflows.