Abstract
The pharmaceutical industry faces growing pressure to develop innovative, affordable products faster. Completing clinical trials on time is crucial, as revenue strongly depends on the finite patent protection. In this paper, we consider dynamic resource allocation for pharmaceutical product portfolio management and clinical trial scheduling, proposing a modelling framework, where resource profiles for ongoing clinical trials are flexible, with the possibility to add additional resources, thereby accelerating the completion of a clinical trial and enhancing pipeline profitability. Specifically, we treat both resource profiles and clinical trial scheduling as decision variables in the management of multiple pharmaceutical products to maximise the expected discounted profit, accounting for uncertainty in clinical trial outcomes. We formulate this problem as a Markov decision process and design a Monte Carlo tree search approach that can identify the best decision for each state by utilising a base policy to estimate value functions. We significantly improve the algorithm efficiency by proposing a statistical racing procedure using correlated sampling (common random numbers) and Bernstein's inequality. We demonstrate the effectiveness of the proposed approach on a pharmaceutical drug development pipeline problem, finding that the proposed modelling framework with flexible resource profiles improves the resource efficiency and profitability, and the proposed Monte Carlo tree search algorithm outperforms existing approaches in terms of efficiency and solution quality.
Original language | English |
---|---|
Pages (from-to) | 1-16 |
Number of pages | 16 |
Journal | European Journal of Operational Research |
Early online date | 19 Jan 2025 |
DOIs | |
Publication status | E-pub ahead of print - 19 Jan 2025 |
Keywords / Materials (for Non-textual outputs)
- dynamic programming
- product scheduling
- flexibility
- sampling