关键词: 不均衡数据, 混合模型, 多标记学习, 特征选择

Abstract: Digital meridian instrument, Health Scale of Traditional Chinese medicine and the four diagnostic instruments are the auxiliary diagnostic tools commonly used in TCM, providing a lot of clinical data. That the distribution of data is not balanced, and there are many diagnostic markers in the same case are a common phenomenon in clinical data. The sub-health data are used to explore the classification method of imbalanced data, kidney disease data are used to integrated hybrid classification model of the three kinds of auxiliary diagnostic tools, data of cardiovascular disease, dyslipidemia and uric acid increased disease are used to explore multi-label data classification method. The experiments have achieved good classification effect, and the selected features accord with the medical theory and have clinical significance.

Key words: imbalance data, hybrid models, multi-label learning, feature selection