Salehi Faraz, Mirzapour Al-E-Hashem S Mohammad J, Moattar Husseini S Mohammad, Ghodsypour S Hassan
Department of Industrial Engineering and Management Systems, Amirkabir University of Technology, Hafez Avenue, Tehran, 15875-4413 Iran.
Rennes School of Business, 2 Rue Robert d'Arbrissel, 35065 Rennes, France.
Ann Oper Res. 2023;323(1-2):331-360. doi: 10.1007/s10479-022-05052-0. Epub 2022 Nov 8.
The need for a study of project portfolio optimization in pharmaceutical R&D has become all the more urgent with the outbreak of COVID-19. This study examines a new model for optimizing R&D project portfolios under a decentralized decision-making structure in a pharmaceutical holding company. Specifically, two levels of decision makers hierarchically decide on budget allocation and project portfolio selection-scheduling to maximize their profit, and we formulate the problem as a bi-level multi-follower mixed-integer optimization model. At the upper level, the investment company has complete knowledge of the subsidiaries' response, acts first, and decides on the best budget allocation. At the lower level, each subsidiary responds to the allocated budget and decides on its portfolio scheduling. Since the lower level represents several mixed-integer programming problems, solving the resulting bi-level model is challenging. Therefore, we propose an efficient hybrid solution approach based on parametric optimization and convert the bi-level model into a single-level mixed-integer model. To validate it, we solve a case and discuss the optimal strategy of each actor. The experimental results show that the planned project portfolio for each subsidiary of the holding company is drastically affected by the allocated budget and its decisions.
The online version contains supplementary material available at 10.1007/s10479-022-05052-0.
随着新冠疫情的爆发，对制药研发项目组合优化进行研究的需求变得更加迫切。本研究考察了在一家制药控股公司的分散决策结构下优化研发项目组合的新模型。具体而言，两个层级的决策者分阶段决定预算分配以及项目组合选择与调度，以实现利润最大化，我们将该问题表述为一个双层多跟随者混合整数优化模型。在上层，投资公司完全了解子公司的反应，率先行动并决定最佳预算分配。在下层，每个子公司根据分配的预算做出反应并决定其项目组合调度。由于下层代表几个混合整数规划问题，求解由此产生的双层模型具有挑战性。因此，我们提出一种基于参数优化的高效混合求解方法，并将双层模型转换为单层混合整数模型。为了验证它，我们求解了一个案例并讨论了每个参与者的最优策略。实验结果表明，控股公司每个子公司的计划项目组合受到分配预算及其决策的显著影响。
在线版本包含可在10.1007/s10479-022-05052-0获取的补充材料。
