Date of Award

8-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Industrial Engineering

Committee Chair/Advisor

Kevin M Taaffe

Committee Member

Ronald G Pirrallo

Committee Member

Sudeep Hegde

Committee Member

Tugce Isik

Abstract

Over 151 million people visit US Emergency Departments (EDs) annually. The diverse nature and overwhelming volume of patient visits make the ED one of the most complicated settings in healthcare to study. ED overcrowding is a recognized worldwide public health problem, and its negative impacts include patient safety concerns, increased patient length of stay, medical errors, patients left without being seen, ambulance diversions, and increased health system expenditure. Additionally, ED crowding has been identified as a leading contributor to patient morbidity and mortality. Furthermore, this chaotic working environment affects the well-being of all ED staff through increased frustration, workload, stress, and higher rates of burnout which has a direct impact on patient safety.

This research takes a step-by-step approach to address these issues by first forecasting the daily and hourly patient arrivals, including their Emergency Severity Index (ESI) levels, to an ED utilizing time series forecasting models and machine learning models. Next, we developed an agent-based discrete event simulation model where both patients and physicians are modeled as unique agents for capturing activities representative of ED. Using this model, we develop various physician shift schedules, including restriction policies and overlapping policies, to improve patient safety and patient flow in the ED. Using the number of handoffs as the patient safety metric, which represents the number of patients transferred from one physician to another, patient time in the ED, and throughput for patient flow, we compare the new policies to the current practices. Additionally, using this model, we also compare the current patient assignment algorithm used by the partner ED to a novel approach where physicians determine patient assignment considering their workload, time remaining in their shift, etc.

Further, to identify the optimal physician staffing required for the ED for any given hour of the day, we develop a Mixed Integer Linear Programming (MILP) model where the objective is to minimize the combined cost of physician staffing in the ED, patient waiting time, and handoffs. To develop operations schedules, we surveyed over 70 ED physicians and incorporated their feedback into the MILP model. After developing multiple weekly schedules, these were tested in the validated simulation model to evaluate their efficacy in improving patient safety and patient flow while accounting for the ED staffing budget.

Finally, in the last phase, to comprehend the stress and burnout among attending and resident physicians working in the ED for the shift, we collected over 100 hours of physiological responses from 12 ED physicians along with subjective metrics on stress and burnout during ED shifts. We compared the physiological signals and subjective metrics to comprehend the difference between attending and resident physicians. Further, we developed machine learning models to detect the early onset of stress to assist physicians in decision-making.

Author ORCID Identifier

https://orcid.org/ 0000-0001-5410-9894

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.