Statistical Characterization of Patient Response to Offered Access Delays Using Healthcare Transactional Data

We present a novel framework to characterize the probability that an offered appointment with k-days' access delay will be booked and subsequently attended by a patient. We refer to this probability as the "probability of realization'" of an offered appointment, and demonstrate how empirical characterizations of this measure can be used to identify improved policies for managing patient demand and enabling the most intentional use of clinical care resources. We consider the estimation problem in the context of new patients looking to establish care at a clinic, and offer a model of patient responses during an appointment scheduling encounter with an agent. We define different cases of data availability and demonstrate the effectiveness of framework in each case to ensure generalizability. We first demonstrate the accuracy of estimations using simulated data, and then highlight behavioral differences between different patient groups using real-life transactional data from two clinical departments. Finally, we clearly demonstrate a practical use case for the obtained realization probabilities by showing that they can be reliably used as input to a novel time windows-based patient prioritization protocol that allows effective management of demand from different classes by explicitly considering wait sensitivities of patients