Gaussian Process Regression-Based Bump Hunting in Future Searches for the X17 Boson

Recent experiments by the ATOMKI collaboration have observed a hypothetical boson with mass 17 MeV, commonly referred to as X17, which could provide insight into BSM physics. Independent experiments are now being conducted to test this claim. A central challenge for new particle searches is reliably modeling the background in invariant mass spectra, where conventional functional forms can introduce bias or reduce sensitivity around local excesses. We present a Gaussian process regression-based bump hunting framework designed to address these challenges. Gaussian processes, in contrast to functional forms, provide a nonparametric approach to background estimation with the ability to quantify uncertainty. We demonstrate this new technique with preliminary figures from an independent study of a future JLab X17 experiment.