Discovery and Validation of Traditional Chinese and Western Medicine Combination Antirheumatoid Arthritis Drugs Based on Machine Learning (Random Forest Model)

Abstract

The combination of traditional Chinese medicine (TCM) and Western medicine is a promising method for treating rheumatoid arthritis (RA). Combining the two fully exploits the advantages of Western and TCM to treat RA and has the potential to greatly improve the therapeutic effect on RA. In this study, we developed a combination drug training set by using 16 characteristic variables based on the characteristics of small molecules of TCM ingredients and Food and Drug Administration-certified combination drug data downloaded from the DrugCombDB database. Furthermore, we compared the prediction and classification abilities of five models: the $k$ -nearest neighbors, naive Bayes, support vector machine, random forest, and AdaBoost algorithms. The random forest model was selected as the classification and prediction model for Western and TCM and Western combination drugs. We collected data for 41 small molecules of TCM ingredients from the Traditional Chinese Medicine Systems Pharmacology database and 10 small molecule drugs commonly used in anti-RA treatment from the DrugBank database. Combinations of Western and TCM for anti-RA treatment were screened. Finally, the CellTiter-Glo method was used to determine the synergy of these combinations, and the 15 most predicted drug combinations were carried out experimental verification. Myricetin, rhein, nobiletin, and fisetin had high synergy with celecoxib, and rhein had high synergy with hydroxychloroquine. The preliminary findings of this study can be further applied for practical clinical anti-RA combined treatment strategies and serve as a reference for clinical treatment of RA with integrated Western and TCM.