中文摘要：

针对中医临床在对有病个体分类时获得的数据很可能是不均衡的、缺失的,而且这种数据往往偏向于无病的个体这一问题,采用病例匹配最近邻填充算法（P-KNN算法）填充缺失值,同时使用基于风险预报准则特征选择的非对称Bagging（PRIFEAB）算法处理睡眠情绪类疾病的不均衡数据和特征选择问题。在中医临床采集的经络电阻数据上的实验结果显示,PRIFEAB算法改善了曲线下面积,并且选择的特征也符合中医学相关理论,同时P-KNN算法比平均值填充算法具有更好的性能。

英文摘要：

In regard of classification of the traditional Chinses medicine （TCM） clinical data, the clinical data was likely to be imbalanced and had missing values. And this data tended to be more biased in favor of disease-free individuals. A missing value filling algorithm （Patient-KNN, P-KNN） was proposed to deal with the issue of missing values, using the prediction risk based feature election for asymmetric bagging （PRIFEAB） algorithm to deal with the imbalanced and fea- ture selection in TCM clinical data of sleep and emotion disorder diseases. The experimental results on TCM clinical data collection of meridian resistance show that the PRIFEAB algorithm improves the AUC and the selected features are also in line with the traditional Chinese medicine theory. P-KNN performs better than using the average values filling algorithm.