4 Presentations

• 
  10:30 AM - 10:55 AM

  CANCELED : Modelling Trade-offs in Efficiency, Equity, and Fairness in Public Defibrillator Placement

  • Benjamin Leung, presenter, University of Toronto
  • Gareth Clegg, The University of Edinburgh
  • Diane Lac, The University of Edinburgh
  • Timothy C.Y. Chan, University of Toronto

  The maximum coverage location problem (MCLP) is a useful approach to determine optimal locations to publicly accessible defibrillators for out-of-hospital cardiac arrest, and has been shown to outperform population-guided heuristics and clinical guidelines. Prior research has focused on maximizing spatial coverage of cardiac arrests across the whole study region; however, this may lead to allocations that are inequitable or unfair across communities of varying geographies, demographics, and socioeconomic levels. We introduce formulations that incorporate trade-offs between the efficiency, equity, and fairness of coverage across study subregions, and compare the standard MCLP with our proposed formulation using cardiac arrest and defibrillator location data from Scotland.
• 
  10:55 AM - 11:20 AM

  Locating helicopter ambulance bases in Iceland: Efficient and fair solutions

  • Armann Ingolfsson, presenter, University of Alberta
  • Björn Gunnarsson, University of Akureyri
  • Kristrún María Björnsdóttir, University of Akureyri
  • Sveinbjörn Dúason, University of Akureyri

  The aim of this study was to find efficient and fair helicopter ambulance base solutions for Iceland. We used population data from Statistics Iceland and incident data from the National Emergency Dispatch Center to estimate service demand. Base locations and coverage for different response time thresholds were determined using two optimization models, in greenfield scenarios and conditioned on an existing base in Reykjavik. One model (the maximal coverage location problem, MCLP) aimed for maximal coverage of demand, whereas the other also penalized uncovered (that is, beyond time threshold) demand. Base locations were usually the same for both models when one and two bases were sited. It is possible to cover 84% of demand within 60 minutes from the existing base in Reykjavik if the activation time is shortened to 15 minutes. Locating another base in Akureyri increases coverage to 95% with a 60-minute response time threshold.
• 
  11:20 AM - 11:45 AM

  Joint Appointment and Reentry Scheduling: Mitigating Onsite Overcrowding in Outpatient Services

  • Yichuan Ding, presenter, McGill University

  In this paper, we study outpatient scheduling problem in the presence of both walk-in arrivals and patient reentry. The model is based on the new growing care coordination model -- the integrated practice unit -- that have physician consults, imaging facilities, and other services co-located in the same place, such that patients can finish a sequence of visits within a day. This business model facilitates patient care and improves patient experiences and outcomes, but presents unique challenges in scheduling and patient flow management. We develop a novel, iterative algorithm to efficiently solve the joint appointment time and reentry time scheduling problem, with provable bounds. Our algorithm also deals with possible endogeneity between the schedule and system parameters, which is often overlooked in the literature. Collaborating with a large teaching hospital in China, we demonstrate that our algorithm can reduce patient delay and mitigate overcrowding in comparison to algorithms that ignore patient reentry. In particular, it reduces the total cost by $13\%\sim 18\%$ compared to the current policy used in our collaborative organization. Meanwhile, the simplicity and adaptability of our algorithm allows implementation -- an ongoing effort we have with our healthcare partner.
• 
  11:45 AM - 12:10 PM

  Guiding Physicians with Time-dependent Patient Selection Policies

  • Mahdi Shakeri, presenter,
  • Marco Bijvank, University of Calgary

  Physicians in emergency departments (EDs) have their own discretion to select the next patient to be seen. Since they operate under an increased workload, such personalized decisions may lead to practices that are less than optimal from a resource utilization perspective and can negatively impact operational and clinical outcomes. In this work, we derive patient selection strategies to guide ED physicians. This is a complex decision and requires the consideration of several factors such as patients’ severity scores, wait times, and the presence of returning patients. In particular, we present a time-dependent policy where the time remaining in a physician’s shift plays an important role. In particular, any patients who are still under the care of the physician at the end of the shift must be transferred to another physician (i.e., patient hand-offs). This is a practice known to compromise the quality of patient care. We formulate an optimal control problem by considering a cost function that captures patient wait times, their acuity, and patient hand-offs. Numerical experiments demonstrate that our proposed time-dependent patient selection policy significantly reduces patient hand-offs compared to traditional time-independent policies while maintaining comparable waiting times and length-of-stay durations.