Josiah Poon1, Zhe Luo, Run-shun Zhang. 1. School of Information Technologies, University of Sydney, Australia. josiah.poon@sydney.edu.au
Abstract
OBJECTIVE: To find an appropriate feature representation in the biclustering of symptom-herb relationship in Chinese medicine (CM). METHODS: Four different representation schemes were tested in identifying the complex relationship between symptoms and herbs using a biclustering algorithm on an insomnia data set. These representation schemes were effective count, binary value, relative success ratio, or modified relative success ratio. The comparison of the schemes was made on the number and size of biclusters with respect to different threshold values. RESULTS AND CONCLUSIONS: The modified relative success ratio scheme was the most appropriate feature representation among the four tested. Some of the biclusters selected from this representation scheme were known to follow the therapeutic principles of CM, while others may offer clues for further clinical investigations.
OBJECTIVE: To find an appropriate feature representation in the biclustering of symptom-herb relationship in Chinese medicine (CM). METHODS: Four different representation schemes were tested in identifying the complex relationship between symptoms and herbs using a biclustering algorithm on an insomnia data set. These representation schemes were effective count, binary value, relative success ratio, or modified relative success ratio. The comparison of the schemes was made on the number and size of biclusters with respect to different threshold values. RESULTS AND CONCLUSIONS: The modified relative success ratio scheme was the most appropriate feature representation among the four tested. Some of the biclusters selected from this representation scheme were known to follow the therapeutic principles of CM, while others may offer clues for further clinical investigations.
Authors: Jinpeng Chen; Josiah Poon; Simon K Poon; Ling Xu; Daniel M Y Sze Journal: Evid Based Complement Alternat Med Date: 2015-06-08 Impact factor: 2.629