Monday, 27 June 2016

Detection of consensuses and treatment principles of diabetic nephropathy in traditional Chinese medicine: A new approach

Open Access

Detection of consensuses and treatment principles of diabetic nephropathy in traditional Chinese medicine: A new approach

Open Access funded by Beijing University of Chinese Medicine
Under a Creative Commons license



To propose and test a new approach based on community detection in the field of social computing for uncovering consensuses and treatment principles in traditional Chinese medicine (TCM).


Three Chinese databases (CNKI, VIP, and Wan Fang Data) were searched for published articles on TCM treatment of diabetic nephropathy (DN) from their inception until September 31, 2014. Zheng classification and herb data were extracted from included articles and used to construct a Zheng classification and treatment of diabetic nephropathy (DNZCT) network with nodes denoting Zhengs and herbs and edges denoting corresponding treating relationships among them. Community detection was applied to the DNZCT and detected community structures were analyzed.


A network of 201 nodes and 743 edges were constructed and six communities were detected. Nodes clustered in the same community captured the same semantic topic; different communities had unique characteristics, and indicated different treatment principles. Large communities usually represented similar points of view or consensuses on common Zheng diagnoses and herb prescriptions; small communities might help to indicate unusual Zhengs and herbs.


The results suggest that the community detection-based approach is useful and feasible for uncovering consensuses and treatment principles of DN treatment in TCM, and could be used to address other similar problems in TCM.


  • Social computing;
  • Community detection;
  • Zheng classification and treatment;
  • Diabetic nephropathy;
  • Traditional Chinese medicine


In complex network theory, a social network is a social structure made of nodes (individuals or organizations) connected by edges to exemplify various relationships such as friendship, affiliation, or cooperation. Community detection is one of the fundamental tasks in social network analysis. It solves problems by studying groups rather than individuals. Finding a community in a social network is a function of identifying a group of nodes that interact with each other more frequently than with nodes outside the group.1 The real-world significance of identifying such communities are, for example, that friends in the same group share more similar interests and interact with each other more frequently2; that community analysis has uncovered thematic clusters on the Internet3; and, in biochemical or neural networks, that communities may be functional modules.4 Nowadays, social computing research has gradually shifted from its traditional research fields such as computer science and engineering to other fields such as health services and communications.5 He6 et al and Chang7 et al extended social cooperation networks to the field of traditional Chinese medicine (TCM) based on a theory that different herbs work together in a complementary manner to treat a disease. Their research also presented a traditional Chinese herbal prescription formulation network (TCHPFN).
In China, TCM herbal formulas have been widely used to treat many diseases. Zheng classification and treatment (ZCT, bian zheng lun zhi, in Chinese) is a unique feature of TCM, and use of TCM herbal formulas must follow ZCT. Many studies have been conducted to illustrate that treatment based on Zheng classification can improve specificity and efficiency in both TCM and Western medicine. 8, 9 and 10 However, Zheng classification depends mostly on the observations, knowledge, and clinical experience of TCM practitioners.11 Zheng diagnosis always varies from practitioners, leading to different formulas, although with equivalent efficacy. There might be underlying consensuses and treatment principles among TCM practitioners to guide their treatments. Therefore, we proposed a community detection-based approach to uncover the underlying consensuses and treatment principles, and tested the approach by applying it to the Zheng classification and treatment of diabetic nephropathy (DNZCT) data. We present the DNZCT network as a social cooperation network of different Zhengs and herbs, and analyze the potential communities in the DNZCT.