Abstract: All-cause mortality in hemodialysis and peritoneal dialysis patients exhibits modality-specific biological mechanisms, while current risk prediction models face challenges in cross-modality validation and temporal treatment parameter integration. Developing machine learning frameworks that dynamically incorporate multidimensional clinical data and dialysis-specific variables could overcome limitations in traditional model construction, ultimately supporting precision decision-making for end-stage renal disease management.

Key words: Hemodialysis, Peritoneal dialysis, All-cause mortality, Predict model