2582. AI-Driven Drug Shortages Mitigation: A Reinforcement Learning and Optimization Framework for Adaptive Supply Chains
Invited abstract in session WB-39: Sustainable Supply Chains I, stream Sustainable & Resilient Systems and Infrastructures.
Wednesday, 10:30-12:00Room: Newlyn LG.01
Authors (first author is the speaker)
| 1. | Seán McGarraghy
|
| Management Information Systems, University College Dublin | |
| 2. | Ali Ala
|
| Mechanical and Materials Engineering, University College Dublin | |
| 3. | Abu Shakil Ahmed
|
| School of Mechanical and Materials Engineering, University College Dublin | |
| 4. | Vincent Hargaden
|
| Systems Engineering, University College Dublin |
Abstract
Drug shortages continue to be a significant concern in global healthcare supply chains, resulting in delays to patient care, increased operating costs, and bottlenecks in resource distribution. Conventional machine learning algorithms tend to be used to forecast shortages, but this research proposes that optimization combined with predictive analytics would better model disruptions to these supply chains. Therefore, a hybrid framework is proposed which combines Reinforcement Learning (RL) and Optimization to forecast and mitigate the effects of medicine shortages via adaptive supply chain decision-making. The proposed methodology combines Deep Q-Networks (DQN) with Mixed-Integer Linear Programming (MILP) and metaheuristic optimization models to enhance supply chain robustness. The reinforcement learning agent enhances supplier selection, inventory distribution, and real-time procurement tactics, mitigating cost and scarcity threats. The effectiveness of the model is assessed using critical parameters, including shortage reduction rates, cost savings, and decision-making efficiency. The results indicate that the RL-optimization framework is superior to both RL and conventional supply chain models, offering an AI-driven decision support system for regulators, hospitals, and pharmaceutical distributors.
Keywords
- Supply Chain Management
- Artificial Intelligence
- Health Care
Status: accepted
Back to the list of papers