Global warming is causing glaciers to retreat and glacial lakes to expand in the Himalayas, which amplifies the risk of glacial lake outburst debris flows (GLODFs) and poses a significant threat to downstream lives and infrastructures. However, the complex interplay between GLODF occurrences and associated indicators, coupled with the lack of a comprehensive susceptibility indicator system that considers the entire GLODF process, presents a substantial challenge in assessing GLODF susceptibility in the Himalayas. This study proposes a process-driven GLODF susceptibility assessment indicator system responding to climate change that considers the complete process of GLODF formation, incorporating relevant parameters about upstream, themselves, and downstream of glacial lakes. Furthermore, to mitigate subjective factors associated with traditional evaluation methods, we developed three novel hybrid machine-learning models by integrating classic machine-learning algorithms with the whale optimization algorithm (WOA) to delineate the distribution of GLODF susceptibility in the Himalayas. All the hybrid models effectively predicted the GLODFs occurrence, with the WOA-SVC model demonstrating the highest prediction accuracy. Approximately 34% of the catchments exhibit high and very high susceptibility levels, primarily concentrated along the north and south sides of the Himalayan ridge, particularly in the eastern and central Himalayas. Indicators capturing the physical formation process of hazards, such as topographic potential (highest relative importance value of 40%), can precisely identify GLODF. A total of 128 catchments pose potential transboundary threats, with 24 classified as having a very high susceptibility level and 25 as having a high susceptibility level. Notably, the border region between China and Nepal is a prominent hotspot for transboundary threats of GLODF. These findings can provide valuable clues for disaster prevention, mitigation, and cross-border coordination in the Himalayas.