Leveraging machine learning tools to expedite oral modified release formulation development

利用机器学习工具加快口服缓释制剂的开发

• 批准号: 2594361
• 金额: --
• 依托单位: University College London
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2021
• 资助国家: 英国
• 起止时间: 2021 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2594361
• 关键词: Leveraging; machine; learning; tools; expedite

Oral delivery is the most preferred route of administration, accounting for over 90% of the global market share available on the market. The oral delivery market continues to grow, and is expected to reach US 150bn in the coming years. However, behaviour of drugs and delivery systems in the intestine depends on many physiological factors including fluid volume, fluid composition, transit, motility, bacteria and pH, which are further influenced by food, gender and age. These are often considered well understood, but their true variability and idiosyncrasies are not fully appreciated or utilised in intestinal dosage form design or in vitro testing. The project proposes to advance oral delivery through harnessing machine learning (ML) tools to optimise oral product development of modified release biological products. Current research in drug delivery continue to use trial-and-error, which is costly, resource-intensive and time-consuming. Given the economic climate of R&D, empirical approaches are no longer sustainable. There has been a shift in recent years to use simulation and predictive tools to help efficiently accelerate formulation development. Such simulations are performed computational, referred to as in silico modelling, which is helping researchers to minimise the already vast formulation space. ML is one emerging in silico tool that is gaining traction for its ability to outperform humans in decision-making tasks. ML is a subset technology of artificial intelligence (AI), that makes prediction on future outcome from existing data. Increasingly becoming the most important commodity of the 21st century, data can be ubiquitously found throughout pharmaceutics and allied fields. However, compression of vast amount is challenging, and hence, ML is needed. The project will explore different ML strategies to help develop oral dosages with precise control release.The project will involve a mixture of both physical and computational experiments, preparing the student for research in the 21st Century. Aspects of the project will involve:- Assess different data acquisition protocols- Asses the feasibility of ML for small datasets- Develop ML models that can integrate data from different characterisation techniques- Establish an ML pipeline to represent end-2-end drug development- Explore explainable modelling algorithms- Address the gap in the lack of informatics in the pharmaceutics domain- Experimentally validate the ML models

口服给药是最优选的给药途径，占全球市场份额的90%以上。口服给药市场持续增长，预计在未来几年将达到1500亿美元。然而，药物和递送系统在肠道中的行为取决于许多生理因素，包括流体体积、流体组成、运输、运动性、细菌和pH，这些因素还受到食物、性别和年龄的影响。这些通常被认为是很好理解的，但其真正的变异性和特异性并没有完全理解或用于肠剂型设计或体外试验。该项目建议通过利用机器学习（ML）工具来优化口服缓释生物制品的产品开发，以促进口服给药。目前的药物输送研究继续采用试错法，这种方法成本高、资源密集且耗时。鉴于研发的经济气候，经验主义方法不再可持续。近年来，使用模拟和预测工具来帮助有效地加速配方开发已经发生了转变。这种模拟是通过计算机进行的，被称为计算机建模，这有助于研究人员最大限度地减少已经很大的配方空间。机器学习是一种新兴的计算机工具，由于其在决策任务中优于人类的能力而获得了越来越多的关注。ML是人工智能（AI）的一个子集，它可以从现有数据中预测未来的结果。数据日益成为21世纪世纪最重要的商品，在制药和相关领域随处可见。然而，大量的压缩是具有挑战性的，因此需要ML。该项目将探索不同的ML策略，以帮助开发具有精确控制释放的口服剂量。该项目将涉及物理和计算实验的混合物，为学生在21世纪的研究做准备。该项目的各个方面将涉及：-评估不同的数据采集协议-评估ML用于小数据集的可行性-开发可以集成来自不同表征技术的数据的ML模型-建立ML管道以代表端到端药物开发-探索可解释的建模算法-解决制药领域缺乏信息学的差距-实验验证ML模型