Reconstructing End-User Demand under Batch-Level Ambiguity: Flow-Constrained Record Linkage and Audit Allocation in Orthopedic Consignment Networks

Abstract This study examines how batch-level records can reveal the commercial provenance of orthopedic implant use and how limited reconciliation resources should be directed toward the most consequential unresolved demand. Methodologically, a flow-constrained probabilistic linkage model connects one month of clinical implantation records to nine months of settlements, adjusting for reporting delays, estimating residual channel states, and comparing audit policies under alternative treatments of repeated records, observation windows, and distributor concentration. Empirical results show that candidate support reaches 42.41% (229/540). Among the 311 unmatched records, the posterior assigns 31.58% to censoring-related reporting lag, 24.57% to batch or capacity competition, and 43.86% to outside-ledger pathways, including wholesale pooling. At a 20% review budget, value-of-information (VoI)-driven auditing captures 67.7% of reducible uncertainty—equivalent to 52.8% of initial uncertainty—outperforming volume-greedy selection by 18.2 percentage points on the initial-uncertainty scale, while conservative duplicate collapse preserves 91.7% of the top-108 audit selection. Ultimately, this research links uncertainty about batch-level provenance to the allocation of audit effort; by combining quantity constraints, reporting delays, and attribution entropy, it demonstrates where digital reconciliation yields the greatest information gain in an asymmetric consignment network.