A data mining approach to selecting herbs with similar efficacy: Targeted selection methods based on medical subject headings (MeSH)

Abstract

Ethno-pharmacological relevanceNatural products have long been the most important source of ingredients in the discovery of new drugs. Moreover, since the Nagoya Protocol, finding alternative herbs with similar efficacy in traditional medicine has become a very important issue. Although random selection is a common method of finding ethno-medicinal herbs of similar efficacy, it proved to be less effective; therefore, this paper proposes a novel targeted selection method using data mining approaches in the MEDLINE database in order to identify and select herbs with a similar degree of efficacy. Materials and methodsFrom among sixteen categories of medical subject headings (MeSH) descriptors, three categories containing terms related to herbal compounds, efficacy, toxicity, and the metabolic process were selected. In order to select herbs of similar efficacy in a targeted way, we adopted the similarity measurement method based on MeSH. In order to evaluate the proposed algorithm, we built up three different validation datasets which contain lists of original herbs and corresponding medicinal herbs of similar efficacy. ResultsThe average area under curve (AUC) of the proposed algorithm was found to be about 500% larger than the random selection method. We found that the proposed algorithm puts more hits at the front of the top-10 list than the random selection method, and precisely discerns the efficacy of the herbs. It was also found that the AUC of the experiments either remained the same or increased slightly in all three validation datasets as the search range was increased. ConclusionThis study reveals and proves that the proposed algorithm is significantly more accurate and efficient in finding alternative herbs of similar efficacy than the random selection method. As such, it is hoped that this approach will be used in diverse applications in the ethno-pharmacology field.