A Measuring Method of Effect Intensity Based on Classified Herbs Regarding Intelligent Mining of Prescription Efficacies

It is an all-important and pivotal technology to translate dosage of each herb in a prescription into effect intensity measured by some models. In this paper, we present relevant Chinese medicine (short in CM) knowledge used for mining of prescription efficacies, and studying necessity of measuring "how effect of any herb depends on dosage" for CM AI systems, and discuss relevant issues and definitions on modeling. In order to measure effect intensity reasonably, a "natural classified herbs method" is presented, which classifies herbs by their dosage range according to usage knowledge of CM and distributing rules of herbs into 38 natural classes, then they are reduce to 6 basic classes. Based on classified herbs, we propose a set of non-linear functions on effect intensity for different classes, which reflects their dosage characteristics. They have been embedded in our intelligent mining system of prescription efficacies (IMSPE), and effects of 560 common herbs have been calculated as well as curative efficacies of about 1000 prescriptions have been mined successfully, which are derived from "ShangHan Lun", "JinGui YaoLue" etc. and modern famous experts. Accuracy of results mined by applying these models has been obviously enhanced.