Joint patient selection and scheduling under no‐shows: Theory and application in proton therapy

We study how to admit and schedule heterogeneous patients by using simple, interpretable, yet effective policies when capacity is scarce, no-show behavior patient- time-dependent, service duration reward are deterministic but patient-dependent, overtime costly. Our work motivated the aforementioned operational challenges that typically face adopters of new technologies in healthcare sector. anchor our on a partnership with proton therapy center Massachusetts General Hospital (MGH), which offers radiation technology for cancer patients. formulate problem as nonlinear integer optimization problem. However, solution this formulation lacks both tractability interpretability, be relevant practice, we limit simple interpretable policies. In particular, propose index-based rule derive analytical performance guarantees it. also calibrate model empirical data from partner hospital, conduct series experiments evaluate proposed policy under practical circumstances. The numerical demonstrate (a) strong policies, (b) their robustness various considerations (e.g., potential misspecification probabilities). results show policy, despite being rule, capable improving about 20% at an organization such MGH, delivering not far optimal across wide range parameters might vary between organizations. This suggests can viewed “one-fits-all” allocation used variety environments no-shows costs need navigated rules.