Date of Graduation

5-2023

Document Type

Thesis

Degree Name

Master of Science in Industrial Engineering (MSIE)

Degree Level

Graduate

Department

Industrial Engineering

Advisor/Mentor

Sandra Eksioglu

Committee Member

Burak Eksioglu

Second Committee Member

Chase Rainwater

Keywords

COVID-19;Healthcare Supply Chain;Inventory Management;LSTM;Machine Learning;Two Stage Stochastic Programming

Abstract

Effective supply chain management is critical to operations in various industries, including healthcare. Demand prediction and inventory management are essential parts of healthcare supply chain management for ensuring optimal patient outcomes, controlling costs, and minimizing waste. The advances in data analytics and technology have enabled many sophisticated approaches to demand forecasting and inventory control. This study aims to leverage these advancements to accurately predict demand and manage the inventory of surgical supplies to reduce costs and provide better services to patients. In order to achieve this objective, a Long Short-Term Memory (LSTM) model is developed to predict the demand for commonly used surgical supplies. Moreover, the volume of scheduled surgeries influences the demand for certain surgical supplies. Hence, another LSTM model is adopted from the literature to forecast surgical case volumes and predict the procedure-specific surgical supplies. A few new features are incorporated into the adopted model to account for the variations in the surgical case volumes caused by COVID-19 in 2020. This study then develops a multi-item capacitated dynamic lot-sizing replenishment model using Mixed Integer Programming (MIP). However, forecasting is always considered inaccurate, and demand is hardly deterministic in the real world. Therefore, a Two-Stage Stochastic Programming (TSSP) model is developed to address these issues. Experimental results demonstrate that the TSSP model provides an additional benefit of $2,328.304 over the MIP model.

Download

Share

COinS