Machine learning-driven optimization of extraction process and development of quality standards for traditional Chinese medicine (TCM) formulae in primary liver cancer

Abstract

Traditional Chinese medicine (TCM) formula extraction optimization is vital for clinical efficacy and standardization. This study targeted an anti-hepatocarcinoma formula, combining orthogonal experimental design (OED) with machine learning (ML) to optimize extraction—focused on extraction yield and paeoniflorin content. OED revealed extraction time as the key factor influencing both metrics, while ML modeling identified optimal parameters. Experimental validation achieved a 43.21% extraction yield and 74.2 mg total paeoniflorin, confirming ML's utility in process refinement. The OED–ML integration proves a powerful tool for TCM preparation optimization, accelerating cost-effective, eco-friendly technology development and advancing formula standardization. This work highlights AI's role in modernizing TCM R&D, offering a replicable framework to balance efficacy, affordability, and sustainability.