CKGROUND: DRG-systems are used to allocate resources fairly to hospitals based on their performance. Statistically, this allocation is based on simple rules that can be modeled with regression trees. However, the resulting models often have to be adjusted manually to be medically reasonable and ethical.METHODS: Despite the possibility of manual, performance degenerating adaptations of the original model, alternative trees are systematically searched. The bootstrap-based method bumping is used to build diverse and accurate regression tree models for DRG-systems. A two-step model selection approach is proposed. First, a reasonable model complexity is chosen, based on statistical, medical and economical considerations. Second, a medically meaningful and accurate model is selected. An analysis of 8 data-sets from Austrian DRG-data is conducted and evaluated based on the possibility to produce diverse and accurate models for predefined tree complexities.RESULTS: The best bootstrap-based trees offer increased predictive accuracy compared to the trees built by the CART algorithm. The analysis demonstrates that even for very small tree sizes, diverse models can be constructed being equally or even more accurate than the single model built by the standard CART algorithm.CONCLUSIONS: Bumping is a powerful tool to construct diverse and accurate regression trees, to be used as candidate models for DRG-systems. Furthermore, Bumping and the proposed model selection approach are also applicable to other medical decision and prognosis tasks.

ckground drgsystem use alloc resourc fair hospit base perform statist alloc base simpl rule can model regress tree howev result model often adjust manual medic reason ethicalmethod despit possibl manual perform degener adapt origin model altern tree systemat search bootstrapbas method bump use build divers accur regress tree model drgsystem twostep model select approach propos first reason model complex chosen base statist medic econom consider second medic meaning accur model select analysi dataset austrian drgdata conduct evalu base possibl produc divers accur model predefin tree complexitiesresult best bootstrapbas tree offer increas predict accuraci compar tree built cart algorithm analysi demonstr even small tree size divers model can construct equal even accur singl model built standard cart algorithmconclus bump power tool construct divers accur regress tree use candid model drgsystem furthermor bump propos model select approach also applic medic decis prognosi task