Performance Guarantees and Optimal Purification Decisions for Engineered Proteins
针对生物制药中蛋白质纯化操作，建立马尔可夫决策模型来平衡产量与纯度、处理原料不确定性，并在Aldevron公司实施后使总前置时间平均降低25%、成本降低20%。
We investigate protein purification operations conducted by biomanufacturers and pharmaceutical companies as part of their research and development efforts. Purification of these proteins involves unique challenges such as balancing the yield and purity trade-offs, dealing with uncertainty in the starting material, and estimating the impact of several interlinked decisions. We develop a Markov decision model and partition the state space into decision zones that provide managerial insights to optimize purification operations. We develop practical guidelines to quantify financial risks, and we characterize the optimal operating decisions based on specific production requirements. The optimization framework has been implemented at Aldevron, a contract biomanufacturer specializing in proteins, and has resulted in 25% reduction in the total lead times and 20% reduction in the costs of protein purification operations on average.