Graduate Student Research Seminar Day ‑ Apr 7, 2021

You are cordially invited to the Graduate Student Research Seminar of the Department of Industrial Engineering

Date: Wednesday, April 7, 2021
Time: 1:00 - 3:30 PM
Venue: Online Event

 

Schedule:

1:00-1:30 PM Cecil Ash
Distributionally robust optimization of healthcare supply chains under risks of disruption from the COVID-19 Pandemic

1:30-2:00 PM Melissa Gillis
A simulation-optimization framework for response policies to minimize infections under budget constraints

2:00-2:30 PM Alireza Forouzangohar
Maritime SAR vessel location-allocation with effectiveness measures

2:30-3:00 PM Natalie Ash
Post COVID-19 patient throughput simulation for surgical resource allocation

3:00-3:30 PM Jack Campbell
Siting primary care clinics to meet daytime and afterhours objectives

 

Abstracts:

Distributionally robust optimization of healthcare supply chains under risks of disruption from the COVID-19 pandemic
Cecil Ash, MASc. Student

The COVID-19 pandemic has struck health service providers around the world with dire shortages, inflated prices, and volatile demand of personal protective equipment (PPE). This demonstrates the importance of supply chain resilience which is the ability to withstand and recover from disruptions to the supply chain. Our research explores supply chain resilience in the context of the Nova Scotia Health Authority (NSHA) procuring PPE during the COVID-19 pandemic. A multi-period multi-objective mixed-integer programming model is presented for PPE supply planning under uncertainty in the supply, price, and demand of PPE caused by pandemics. The first objective function, cost minimization, is formulated using stochastic, robust, and distributionally robust optimization. The second objective function, service-level, follows minimax robustness by minimizing the maximum shortage of any product in any time period. The epsilon-constraint method is used to generate sets of Pareto-optimal solutions and analyze the trade-off between these two competing objectives. Numerical experiments are conducted to analyze the efficacy of emergency inventory and increased inventory levels as risk mitigation strategies. We consider uncertainty scenarios based on actual pandemic trajectories seen around the world during the COVID-19 pandemic including single-wave, two-wave, and exponential growth.

A simulation-optimization framework for response policies to minimize infections under budget constraints
Melissa Gillis, MASc. Student

Epidemics require dynamic response strategies that encompass a multitude of policy alternatives and that balance health, economic and societal considerations. We propose a simulation-optimization framework to aid policymakers select closure, protection, and travel policies to minimize the total number of infections under a limited budget. The proposed framework combines a modified, age-stratified SEIR compartmental model to evaluate the health impact of response strategies and a Genetic Algorithm to effectively search for better strategies. We implemented our framework on a real case study in Nova Scotia to devise optimized response strategies to COVID-19 under different budget scenarios and found a clear trade-off between health and economic considerations. Closure policies seem to be the most sensitive to policy restrictions, followed by travel policies. On the other hand, results suggest that practicing social distancing and wearing masks are necessary under all scenarios. The framework is generic and can be extended to encompass vaccination policies and to use different epidemiological models and optimization methods.

Maritime SAR vessel location-allocation with effectiveness measures
Alireza Forouzangohar, MASc. Student

Search and Rescue (SAR) continues to be a significant issue in today’s world. The extent and importance of SAR have made it an important operational goal for many countries and an area of focus for many researchers. The aim of SAR is to minimize loss of life, injury, property damage, and environmental risks. Due to the fact that SAR operations consume a considerable amount of resources (such as money, time, and human resources), the response actions should be well-planned and organized. SAR can be defined in many categories like marine, mountain, earthquake, etc. and uncertainty is an inseparable part of these problems, because time, location, and circumstances of the incident are not easily predictable. In this research, we propose a location-allocation model for optimizing maritime SAR vessels in Atlantic Canada. Our model is a non-linear multi-objective model which optimizes access time, cost, effectiveness, and also minimizes failure probability of vessel dispatch to demand grids. In the proposed model, a stochastic approach has been used to address the uncertainty, and historical data has been utilized to obtain the probability distribution of incidents for each demand grid. Different sorts of incidents have been considered to make the problem more realistic. Furthermore, rating scales have been defined to show the capability of each vessel for responding to each sort of incident. Dynamic programming along with metaheuristic approaches are used to solve this problem.

Post COVID-19 patient throughput simulation for surgical resource allocation
Natalie Ash, MASc. Student

In response to the COVID-19 pandemic, most elective surgical procedures in Nova Scotia were cancelled resulting in increased patient waitlist volumes. Understanding the impacts of resource allocation is crucial to developing an effective strategy to reduce the waitlist. A discrete event simulation model was developed to analyze the overall potential throughput of elective surgical patients. Descriptive analytics of two years (2018-2020) of surgery data from the Central Zone of Nova Scotia Health informed the development and inputs of the model. The model facilitated scenario analysis of the impacts of resource allocation. Scenarios for COVID-19 recovery included increasing bed capacity and operating room (OR) hours. Further, COVID-19 scenarios included a second wave, increased OR sanitation time, increased demand, and a combination of the interventions.

The base model, which reflected the current system parameters, indicated the waitlist grew continuously with orthopedics, general surgery, and urology comprising 68%. The outpatient waitlist decreased to a steady state, whereas the inpatient waitlist continuously increased. The scenario analysis quantified that increasing the OR hours and the number of outpatients on the waitlist impacted the waitlist more than increasing the bed capacity. The largest waitlist decrease, 33% as compared to the base model, occurred when ORs were used for two hours of overtime.  The overall throughput is limited by the patient’s surgical specialty as well as the number of ORs available. The number of available OR hours and the types of patients on the surgical waitlist had the largest impact on the patient throughput. These two approaches to resource allocation would positively impact the waitlist created by the COVID-19 pandemic.

Siting primary care clinics to meet daytime and afterhours objectives
Jack Campbell, MASc. Student

Location science is used to determine the optimal geographical placement of primary care resources with operations research models. In determining the optimal placement, we account for the objectives of both patients and physicians. Patients prefer to be close to clinics to ensure access and physicians typically prefer to have minimum panel sizes to ensure consistent appointments. These objectives and the methods used to address them differ between daytime and afterhours settings. Three approaches are considered to address both time settings: independent, sequential, and simultaneous. The independent approach is based on the p-Median problem, and the other two approaches use modified forms of the p-Median. The models are generalized and applied to census data from Nova Scotia. Three case studies are examined using Canadian census data from Halifax, Cape Breton, and modified data from Cape Breton.

 

Contact Person:
Prof. Dr. Floris Goerlandt
email: floris.goerlandt@dal.ca