罕见难治性疾病中医有效治疗方案评价的贝叶斯动态伞式设计方法研究

罕见、难治性疾病病例稀少、治疗复杂、缺乏有效对照导致其不适合传统的固定的临床设计，迫切需要灵活高效的新型设计。本研究提出贝叶斯动态伞式设计，可同时或随时开展多个不同子试验，灵活调整治疗方案，充分利用前期数据进行期中分析及决策。本研究首先对贝叶斯动态伞式设计关键方法-贝叶斯响应动态随机化、贝叶斯层次模型进行理论和模拟研究，确定合适的算法和模型。然后，以特发性肺纤维化（IPF）为范例设计贝叶斯动态伞式试验，在相同主体方案和试验管理机制下，根据证型设立子试验评价其治疗方案，采用贝叶斯响应动态随机化进行随机分配；贝叶斯层次模型进行期中分析，预测后期研究成功概率决定是否继续研究。最终形成可行的贝叶斯动态伞式设计方法及IPF评价平台（如成熟的主体研究方案、方法学、试验管理机制等），有利于集中和共享病例，为后续新的治疗方案评价提供条件，促进IPF领域临床研究，为其他类似疾病提供思路和范例。

Rare patients, complex treatments and lack of effective control are the characteristics of rare, refractory diseases. Therefore the traditional fixed clinical design is not the good choice to assess the efficacy of rare, refractory diseases. It urgently needs flexible and efficient new design. A dynamic umbrella design based on Bayes theory is proposed, which can carry out multiple sub-studies at the same time or at any time, flexibly adjust treatments, and make full use of history data for interim analysis and decision-making. Firstly, the key methodologies of Bayes dynamic umbrella design (Bayesian response dynamic randomization, Bayesian hierarchical model) are studied to determine the appropriate parameters, algorithms and models. Secondly, the Bayesian dynamic umbrella trials will be designed for idiopathic pulmonary fibrosis (IPF) . Under the same main protocol and logistic management measurements, sub-trials were established to evaluate the treatment according to syndrome patterns of TCM, and the Bayesian response randomization will be used. The Bayesian hierarchical model will be used to interim analysis and predict the success probability of the later study and whether to continue the study. Through this study, we will set up a feasible Bayesian dynamic umbrella design methodology and IPF evaluation technology platform (such as main protocol, methodologies, logistic management measurements, etc.) , which will help to centralize and share IPF patients, assess the efficacy of new treatments, promote clinical research in the whole field of IPF, and provide ideas and examples for other similar diseases.