机器学习在半月板骨骼芯片中药物安全的评价

Evaluation of Drug Safety in Meniscus Bone-on-a-Chip by Machine Learning

摘要: 半月板损伤普遍存在于年轻人群中，常规治疗手段是半月板切除术，这容易带来骨关节炎等副作用。组织粘合剂被视为治疗半月板损伤的新型治疗方法，但是可能作为组织粘合剂的生物质材料有很多，影响组织粘合剂作用的参数也多种多样，这在一定程度下降低了新型材料的研发速度。骨骼芯片技术是在体外搭建一个适合组织分化的微环境，它能够有效地辅助药物研发。机器学习作为获取数据信息的重要手段之一，在许多方面都体现出了研究潜力，本文从几个重要算法出发，研究了机器学习在组织粘合剂筛选方面的可行性。我们试图寻找一种方法，它能够将从体外模型中获得的数据信息的背后价值挖掘出来，这对半月板骨骼芯片中药物安全的评价是有意义的。

Abstract: Meniscus injury is common in young people. Meniscectomy is the conventional treatment, which is easy to bring side effects such as osteoarthritis. Tissue adhesive is regarded as a new treatment for meniscus injury, but there are many biomaterials that may be used as tissue adhesive, and the parameters that affect the function of tissue adhesive are also varied, which reduces the development speed of new materials to a certain extent. Bone chip technology is to build a microenvironment suitable for tissue differentiation in vitro, which can effectively assist drug development. Machine learning, as one of the important means to obtain data information, has shown research potential in many aspects. This paper studies the feasibility of machine learning in tissue adhesive screening from several important algorithms. We are trying to find a method that can extract the value behind the data obtained from in vitro model, which is meaningful for the evaluation of drug safety in meniscus bone chip.