Harnessing Artificial Intelligence to Expedite Identification of Therapeutic Phytochemical Combination for Alcoholic Hepatic Injury

Abstract

AbstractPlant‐originated compounds, also known as phytochemicals, could improve health through either daily consumption or medical treatments. Conventional developments of phytochemical therapies are often based on the action mechanism of single compounds. However, compounds within a plant or across different plants are numerous, and their interactions potentially mold its therapeutic effect, which is hard to explore with conventional approaches. To address such challenges, an Artificial Intelligence (AI) platform featuring Monte Carlo method‐based modeling (MCM2) for rapid identification of the optimal dose combinations is devised. MCM2 utilizes both the mean and variances within multiple replicates, and thus maximizes the optimization efficiency. MCM2 is applied to optimize the therapeutic effect on alcoholic hepatic injury of a quaternary dose combination of products from Gynostemma pentaphyllum and Lotus leaf, which are identified through in vitro screening of 50 products extracted from 24 herbal medicine candidates. The optimal combination identified by MCM2 is then tested on murine model. Both in vitro and in vivo results confirmed the AI‐optimized phytochemical combination significantly reduced acute alcoholic liver injury. The AI platform introduced in this study could serve to promptly identify optimal combinatorial regimens with potential synergy, which enhances the efficiency during the development of compound medicine.