Nowadays, China’s traditional herb medicine have been widely recognized by more and more doctors, medical institutions and individuals in western countries. Especially in recent decades, various effective pure plant drugs extracted based on the experience of China’s traditional herb medicine have further consolidated the importance of traditional medicine in the medical field. The research on artificial intelligence methods based on deep learning, on the other hand, have made great breakthroughs in various fields including medical application. However, in traditional medicine, especially the diagnosis and treatment of intelligent Traditional Chinese Medicine (TCM), it is still in its infancy.

Recently, ZHENG Zeyu and his research team at Shenyang Institute of Automation of Chinese Academy of Sciences have proposed an adaptive Traditional Chinese Medicine prescription generation model for intelligent Traditional Chinese Medicine research. The study was published in IEEE ACCESS titled AttentiveHerb: A Novel Method for Traditional Medicine Prescription Generation. 

This method is based on the core idea of TCM diagnosis and treatment process-"syndrome differentiation and treatment". First, the features of the patient’s symptom information are extracted by using deep recurrent neural network. 

Considering the severity of different symptoms of the patient, each herb in the prescription can ben dynamically generated according to the corresponding symptoms, realizing a process of intelligent automatic diagnosis and treatment of TCM. Compared with previous study on intelligent TCM, this method is completely based on the learning method and does not need to manually design any rules by hands in advance, and shows good adaptability. 

Apart from being a new exploration and breakthrough in the research of intelligent traditional Chinese medicine, this method can also inspire future research of applying artificial intelligence technology in outpatient diagnosis and treatment.

The Traditional Chinese Medicine prescription generation model based on deep recurrent neural network (Image by LIU Zhi)