针刺疗效受机体状态的显著影响，基于客观、明确的影响特征科学评价针灸治疗适宜人群对提高临床疗效十分必要。本项目在前期研究的基础上，以脑-肠轴功能失衡的典型疾病——功能性消化不良（FD）为研究对象，采用机器学习领域具有代表性的集成学习方法——随机森林算法，基于症状-情绪-胃肠激素-脑功能和结构异常四者在FD病理变化和针刺干预中的重要性和相关性，融合神经影像表征、临床症状特征、胃肠激素水平、情绪心理状态等多维度数据建立分类预测模型，用于患者的疗效预判；并在新的独立样本中验证模型的精准度，明确针刺治疗FD的重要显效特征；再以其中更为复杂和客观的神经影像学重要特征为感兴趣特征，从纵向研究角度再次验证模型和显效特征在评价针刺胃肠调节适宜人群中的精准度。本研究是引入机器学习方法评价针刺治疗适宜人群以助力精准治疗、提高针灸临床疗效的有益尝试，具有显著的临床实用价值和方法学借鉴意义。

The acupuncture efficacy is significantly impacted by the physical condition of individual. To improve the clinical efficacy, it is necessary to evaluate the positive responsers to acupuncture treatment objectively after identifying the influence factors. Based on our previous studies, this project will take Functional Dyspepsia (FD, the typical functional gastrointestinal disorder induced by the imbalance of brain-gut axis) as the object, and use Random Forest method (RF, the representative ensemble learning method in the machine learning field) as main algorithm, and combine the multidimensional data including the neuroimaging data, clinical variables, gastrointestinal hormone levels and emotional status data to establish the classification/prediction model for acupuncture efficacy. Then, the model validation will be performed with new independent sample, and the valuable features for predication will be extracted. Finally, the valuable neuroimaging features will be selected to revalidate the precision of the model from the longitudinal aspects because they are more objective and complex than other features. This project is a useful attempt to utilize the machine learning method to evaluate the acupuncture for appropriate people, which can not only improve the acupuncture efficacy to assist the precise treatment and provide methodological reference for future studies.