Title:
Improving patient benefits via digital modeling and AI-augmented decision support for supply chains of autologous cell-derived medical products
Improving patient benefits via digital modeling and AI-augmented decision support for supply chains of autologous cell-derived medical products
Author(s)
Tseng, Chin-Yuan
Advisor(s)
Wang, Ben
White, Chelsea
Ma, Xiaoli
White, Chelsea
Ma, Xiaoli
Editor(s)
Collections
Supplementary to
Permanent Link
Abstract
Autologous cell-derived medical product (AuCMP) is a field in regenerative medicine that uses a patient's living cells or tissues to treat diseases. Many AuCMPs have demonstrated promising clinical outcomes in treating various intractable diseases, such as bladder cancer, lymphoma, and myeloma. The US FDA estimated that 10 - 20 AuCMPs will be approved yearly by 2025. Therefore, developing an efficient AuCMP supply chain is essential to meet this unprecedented demand and ensure patients' timely access to life-saving treatment. We envision digital supply chain twin will play an essential role in the future AuCMP design and management. This thesis advances the field of digital supply chain twin by proposing and implementing an AI-augmented simulation-based decision support platform for AuCMP. The research focuses on three topics: i) modeling AuCMP supply chains to improve patient benefits and optimize strategies, including decentralized manufacturing and resource sharing in decentralized networks; ii) incorporating patient/therapy data into manufacturing decision-making; and iii) AI-augmented decision support by integrating reinforcement learning algorithms with simulation. In Chapter 2, the study introduces a novel simulation platform, AuCMP-DSS, designed for AuCMP supply chain decision support. AuCMP-DSS employs agent-based and discrete-event simulation approaches to model the macro- and micro-scale activities in the targeted supply chain. In addition, we proposed a modular approach to enable the model's generalizability for various products. Case studies are conducted to illustrate the platform's utility, particularly in comparing inventory costs and optimizing supply chain capacity. The platform is further used to analyze the trade-offs between centralized and decentralized AuCMP supply chains in the USA context. Chapter 3 explores the importance of real-time patient health data in CAR T-cell therapy manufacturing decisions. By integrating a System Dynamics simulation module into the AuCMP-DSS platform, the research demonstrates the potential benefits of adjusting the manufacturing process based on individual patient health status and prioritizing patients during therapy manufacturing. The potential improvements include enhanced survival rates, increased sales, and greater profits. In Chapter 4, the research confronts two main challenges within decentralized AuCMP manufacturing networks: production job dispatching and capacity planning. Data-driven approaches employing simulation and reinforcement learning algorithms are proposed, and their effectiveness in achieving superior performance under various scenarios is demonstrated. Chapter 5 concludes the thesis by summarizing the main findings and opening up potential avenues for future research. The emerging AuCMP industry and its expected market growth in the near future emphasize the importance of these proposed methodologies and research findings. This dissertation ultimately provides innovative strategies for planning and managing the AuCMP supply chain, facilitating successful commercialization and meet anticipated demand.
Sponsor
Date Issued
2023-07-14
Extent
Resource Type
Text
Resource Subtype
Dissertation