Predictive Scalability in Developing Large Molecule Therapeutics

“数字分子技术”的新兴工具承诺分子开发过程的逐步变化，主要是通过数据生成，数据分析和知识生成的新功能。Lapkin集团在配方和化学工艺开发中的数据分析机器学习（ML）方法方面拥有丰富的专业知识。该项目将使用机械和非线性多变量统计模型的组合，用于大分子治疗剂合成中的决策支持和知识识别。关键的挑战将是开发一个通用的工作流程，该工作流程将灵活地使用ML工具来识别变量的相互依赖性和敏感性。该项目将从探索使用过去的实验数据来生成初始灰色模型的可能性开始，并测试使用过去的一般知识的假设，然后转向开发决策支持工具。

The emerging tools of 'Digital Molecular Technology' promise a step-change transformation of molecule development processes, mainly through new capabilities in data generation, data analysis and knowledge generation. The Lapkin group has developed significant expertise in machine learning (ML) methods for data analysis in formulations and chemical process development. This project will use a combination of mechanistic and non-linear multi-variate statistical models for decision support and knowledge identification in the synthesis of large molecule therapeutics. The key challenge will be the development of a generic workflow that will be flexible in the use of ML tools for the identification of variable interdependencies and sensitivities. The project will start from exploring the possibility of using past experimental data to generate initial grey models, and to test the hypothesis of using past generalised knowledge, then moving to developing decision support tools.