基于复杂网络的糖尿病肾病案例推理证治比较和药效机制研究

本项目在复杂性科学相关理论的指导下，以开放的组学数据库和自主研发的中医药诊疗支撑平台为依托，以案例类比推理建模为路径，选用随机森林和支持向量机等分类器，探索糖尿病肾病（DN）早中晚期证型分类、复方加减；应用ELD和PPAS算法，计算类方间相似度，明确匹配依据。开展Network Meta-Analysis，绘制同一证型下不同干预措施的疗效比对网络，解决治疗方案的择优、排序问题，得出各证型最优方案。开展上述复方用药与效果比较研究，临床层面，可明确中医药防控糖尿病对肾功能损害的机制；基础层面，可揭示复方药效的作用机理。为此，将优选方案结合TCMSP、TTD、STITCH等数据库，进行药靶筛选，预测DN药效成分与潜在靶点的结合度；选用ADME/T计算工具，评估方药利用、吸收、分布、代谢情况，构建起效靶点之间的交互关系与分子网络，从分子水平诠释最优方案药效成分群作用机制。

This research will be performed under the guidance of theories of complexity science, majorly with the method of evidence-based reasoning, and in the support of independently developed TCM Supporting Platform and open source omics databases. Cross testing will be conducted among the classifiers such as random forest and support vector machine to explore the pattern classification of diabetic nephropathy (DN) in early, middle and late stages. In this process, the rule for matching is confirmed using calculated similarities among classified prescriptions with algorithms including ELD and PPAS. Further, a Network Meta-analysis will be carried out, so that the indirect effectiveness comparisons among the classified prescriptions are visualized. From the facet of clinical practice, it correlates with the prevention and control of DN with TCM schemes; and of laboratory research, this may contribute to the discover of mechanism for compound prescriptions. Therefore, we will further make the screening of MS for the most effectives prescriptions base on TCMSP, TTD and STITCH and other database. Then the integration degree between potential target spots for influencing DN and these MSs can be predicted. Using ADME/T kit, the absorption, distribution, metabolism and excretion of the prescriptions will be evaluated to locate the biological pathways of the MSs. Subsequently, we will develop the molecular network of effective proteins.Eventually how the prescriptions takes effect in human body can be figured out on molecular level.