An unsupervised pattern (syndrome in traditional Chinese medicine) discovery algorithm based on association delineated by revised mutual information in chronic renal failure data
An unsupervised pattern (syndrome in traditional Chinese medicine) discovery algorithm based on association delineated by revised mutual information in chronic renal failure data
复制标题
DOI：
10.1142/s0218339007002350
复制
发表时间：
2007-12-01
影响因子：
1.6
通讯作者：
Wang, Wei
中科院分区：
文献类型：
--
作者：
Chen, Jianxin;Xi, Guangcheng;Wang, Wei
证候是中医的基本病理单位和核心概念，根据患者的证候对症用药。然而，很少有研究致力于调查慢性肾衰竭（CRF）患者的综合征数量以及这些综合征是什么。在本文中，我们进行了临床流行病学调查，获得了601例CRF病例，每份报告包括72种症状。基于相互信息描述的关联，我们提出了一种新的模式发现算法来发现可能与中医症状重叠的证候。这里提出了互信息的修订版本，以区分积极和消极关联。该算法自组织发现16种有效模式，每一种模式均由中医医师手动验证，以识别其所属的证候。通过互信息证明了簇的超可加性，并在模型中引入了n类关联概念以降低计算复杂度。该算法的验证是通过使用综合征数据进行的，并在临床上整合为 16 种模式。结果表明，该算法的灵敏度高达96.48%，且每种分类模式均具有临床意义。因此，我们得出结论，该算法为中医背景下的慢性肾衰竭问题提供了极好的解决方案。
The syndrome is the basic pathological unit and the key concept in traditional Chinese medicine (TCM), and the herbal remedy is prescribed according to the syndrome a patient catches. Nevertheless, few studies are dedicated to investigate the number of syndromes in chronic renal failure (CRF) patients and what these syndromes are. In this paper, we carry out a clinical epidemiology survey and obtain 601 CRF cases, including 72 symptoms in each report. Based on association delineated by mutual information, we propose a novel pattern discovery algorithm to discover syndromes, which probably have overlapped symptoms in TCM. A revised version of mutual information is presented here to discriminate positive and negative association. The algorithm self-organizedly discovers 16 effective patterns, each of which is verified manually by TCM physicians to recognize the syndrome it belongs to. The super-additivity of cluster by mutual information is proved and n-class association concept is introduced in our model to reduce computational complexity. Validation of the algorithm is performed by using the syndrome data and consolidated clinically to have 16 patterns. The results indicate that the algorithm achieves a high sensitivity with 96.48% and each classified pattern is of clinical significance. Therefore, we conclude that the algorithm provides an excellent solution to chronic renal failure problem in the context of traditional Chinese medicine.