Study on the medication rule of traditional Chinese medicine in the treatment of acute pancreatitis based on machine learning technology.

Background: To analyze the rule of traditional Chinese medicine in the treatment of acute pancreatitis (AP).

Methods: Using machine learning technology and artificial intelligence, we collected 516 traditional Chinese medicine compounds for treating AP in the recent past 20 years, and analyzed the application of Chinese medicine in the field of AP. The data set was established by the ingredients of each prescription and its corresponding effectiveness. 90% of the data was divided into the training set, and the remaining 10% of the data was used as the test set. We employed random forest method to build a model to predict the efficacy of the prescriptions in the treatment of AP. The R-squared score and mean absolute error was used to evaluate the model's performance.

Results: The most frequently used drugs were rhubarb, Radix Bupleuri, Fructus Aurantii Immaturus, and Mirabilite. Rhubarb and Rhizoma Corydalis had the greatest curative effect. The random forest model that fit all data showed that its R-squared score reached 0.8021. And the results predicted on the test set showed that the R-squared score reached 0.7318.

Conclusions: Soothing the liver, promoting qi, clearing heat, removing obstructions of organs, activating blood, and resolving stagnation are the treatment methods for AP.