Chinese medicine prescription optimization using evolutionary algorithms

Formulating a prescription consisting of functionally synergistic or complementary herbal medicines to treat the syndromes of a given patient is a core therapeutic method in Chinese medicine (CM). However, the process largely depends on the subjective experiences and preferences of CM physicians, and hence can be improved through optimization. In this paper, we formulate the problem of Chinese medicine prescription optimization as a combinatorial optimization problem based on a novel prescription evaluation model, which incorporates CM expert knowledge to comprehensively and quantitatively evaluate herb-syndrome effects, herb-herb interactions, prescription-syndrome effects, and prescription-patient effects. We adapt a collection of popular evolutionary algorithms (EAs) to solve this optimization problem. Our numerical results on a wide set of real-world patient records demonstrate that the EAs can produce personalized prescriptions that have better effects than those created according to the common formulation process of CM physicians.