Modeling Reimbursement-Relevant Drug Data as a Property Graph

Drug-related healthcare data are difficult to integrate because products, substances, indications, classifications, and reimbursement elements are distributed across heterogeneous public sources and linked through overlapping, partly inconsistent relations. This study examines whether a Neo4j-based property graph can support analysis of high-cost drugs in the SwissDRG reimbursement context. Public data from SwissDRG-related material, EMA, Refdata, the Swiss reimbursement environment, and the WHO ATC/DDD system were preprocessed in R and Python and modeled as a labeled property graph. The resulting graph contained more than 27,000 nodes and 41,000 relationships and enabled integrated traversal across products, substances, indications, ATC groups, and reimbursement elements. It supported indication-centered exploration, preserved indirect and source-specific links, and reduced repeated manual reconciliation. The findings suggest that property-graph modeling is well suited to reimbursement-relevant drug data characterized by heterogeneous, weakly aligned, and analytically important relations.