Constructing accurate patient transfer networks between hospitals is critical for understanding the spread of healthcare associated infections through statistical and mathematical modeling, and for determining optimal screening and treatment strategies. The Healthcare Cost & Utilization Project (HCUP) State Inpatient Databases (SID) provide valuable information on patient transfers from publicly obtainable claims databases, yet often give an incomplete picture due to missingness of patient tracking identifiers. We designed a novel imputation algorithm that enabled us to estimate the true number of patient transfers between each pair of hospitals in a state over a specified time period and age group in the presence of these missing identifiers. We then validated the algorithm’s performance through a series of simulation experiments using the HCUP SID, and finally tested the algorithm on multiple states' genuine data. Our proposed method significantly reduced the total mean squared error in predicting the true number of transfers amongst hospitals for all simulation experiments, and it also yielded epidemic simulations that more closely approximated those corresponding to the true patient transfer network.