基于状态辨识的人工智能组方模型设计

Design of Artificial Intelligence Recipe Model Based on State Identification

借助现代计算机技术实现中医智能方药推荐是中医现代化研究的趋势,然而目前中医智能方药推荐模型多以直接选择推荐成方为主,罕见智能组方的文献报道。中医临床实际情况复杂多变,难见“教科书式”的病人,成方难以满足全部需求。因而文章提出一个在状态辨识基础上的智能组方模型,即对状态辨识结果(状态要素集合)进行分组结构化-形成单证-根据单证选用药群-合成处方,并对该模型进行扩展、运用规则对模型进行约束用以解决推荐成方难以满足需求的情况,且更加符合传统中医思维,体现中医全面、整体、动态、个性化的特点。

Recently,there are more and more research interests in designing artificial intelligence recipes of TCM.However,these methods mainly recommend a set recipe for the patient instead of performing an intelligent recipe combination.The clinical situation is complicated,and each patient is showing some difference with the one written in the textbook.Therefore,we cannot directly use a set recipe for all patients.In this study,we proposed an intelligence recipe model based on state identification to deal with these issues.Specifically,we first combine the state syndromes of each patient into different groups(i.e.,disease location-nature,nature)and select corresponding drugs for each group.Meanwhile,we also apply a series of medical constraints to construct the final recipe recommendation.The proposed model is following TCM thinking,which has the characters of comprehensive,holistic,dynamic,and personalized.