3 présentations

• 
  13h30 - 13h55

  EMERGENCY DEPARMENTS LIVE PATIENT LEVEL WAIT TIME PREDICTION IN ONTARIO, CANADA

  • Suting Yang, prés., Data and Decision Sciences, Ontario Health
  • Shabnam Balamchi, Data and Decision Sciences, Ontario Health
  • Saba Vahid, Consulting & Analytics, IQVIA
  • Sophie Foxcroft , System and Infrastructure Planning, Ontario Health
  • Tamer Ahmed, Information Management, Ontario Health
  • Attila Monlar, Product Management & Delivery, Ontario Health
  • Steve Scott, Information Management, Ontario Health

  Majority of patients experience long wait times in emergency departments (EDs), especially lower acuity patients. Considering the availability in live triage data and historical ED wait times, Ontario Health (OH) has the potential to support Ontario hospitals in publishing expected wait times. The objective of this study is to develop a live wait time prediction system utilizing machine learning techniques. We link patient-level ED data at five various-sized hospitals from January 2019 to December 2019 using National Ambulatory Care Reporting System (NACRS) and Electronic Canadian Triage and Acuity Scale (eCTAS). We then assess 9 machine learning methods – multiple linear regression, quantile regression, random forest, support vector regression, elastic net, artificial neural network, gradient boosting, mixed-effect gradient boosted tree, and stacked model – based on appropriate error metrics. The findings demonstrate that the stacked model has superior prediction performance compared to individual models. It decreases the mean absolute error by between 1% and 18% depending on different data sets. The normalized root mean square errors range from 19% to 25%, showing the reliable model fit. This study shows promise for accurate wait time predictions in selected EDs that can benefit operational decision making.
• 
  13h55 - 14h20

  Rationing scarce healthcare capacity: A study of the ventilator allocation guidelines during the COVID-19 pandemic

  • Tolga Aydinliyim, prés., Baruch College, CUNY
  • David Anderson, Villanova University
  • Margret Bjarnadottir, University of Maryland
  • Eren Cil, University of Oregon
  • Michaela Restivo Anderson, Penn Medicine

  We study the existing procedures in the United States for allocating/rationing scarce ventilator capacity and propose alternative (and improved) priority schemes. We first build machine learning models using inpatient records of COVID-19 patients admitted to New York-Presbyterian/Columbia University Irving Medical Center and an affiliated community health center to predict survival probabilities as well as ventilator length-of-use (LOU). Then, we use the resulting point estimators and their uncertainties as inputs for a multiclass priority queueing model with abandonments to assess three priority schemes: (i) SOFA-P (Sequential Organ Failure Assessment based prioritization), which most closely mimics the existing practice by prioritizing patients with sufficiently low SOFA scores; (ii) ISP (incremental-survival-probability), which assigns priority based on patient-level survival predictions; and (iii) ISP-LU (incremental-survival-probability-per-LOU), which takes into account survival predictions and resource use duration. Our findings highlight that ISP-LU achieves a demonstrable improvement over the other two alternatives. Specifically, the expected number of survivals increases and death risk while waiting for ventilator use decreases. We also show that ISP-LU is a robust priority scheme whose implementation yields a Pareto improvement over both SOFA-P and ISP in terms of maximizing saved lives after mechanical ventilation while limiting racial disparity in access to the priority queue.
• 
  14h20 - 14h45

  Reducing waiting time for elective surgeries

  • Alexandru Popp, prés., Concordia University
  • Rustam Vahidov, Concordia University

  The primary goal of this research is to determine if a centralized vs. the present
  decentralized organizational system is beneficial for the patients, in such a manner that the
  priority on the waiting lists of patients is updated to reflect the amount of time they have waited.
  The current study combines quantitative computer simulations (genetic algorithm approach) and qualitative interviews that help in the understanding and interpretation of the simulations’ results.
  The analysis of this research is based on simulating a centralized planning method of elective surgeries planning for five different surgical units in order to provide different perspectives vis-à-vis the current decentralized planning method. The difference between the approaches is provided in order to obtain the best method that minimizes elective surgery waiting time, taking in consideration certain factors. The five departments (neurology, ear-neck-throat, cardio-vascular/thoracic, and ophthalmology) considered for the analysis are important as each unit has different surgical time durations.
  The results of the simulations show that centralization is beneficial for certain surgical
  units and for specific indicators, where decentralization is beneficial for other units and/or
  indicators. However, there is no definitive overall conclusion that centralization is better than
  decentralization or vice-versa.