Screening and validation of Sinomenine hydrochloride deep eutectic solvents ligand based on machine learning algorithm

Zehua Ying, Xuting Guo, Chaoliang Jia, Zhiqi Guan, Qingbo Lv, Wenlong Li,
Screening and validation of Sinomenine hydrochloride deep eutectic solvents ligand based on machine learning algorithm,
Journal of Molecular Liquids,
Volume 439, Part A,
2025,
128934,
ISSN 0167-7322,
https://doi.org/10.1016/j.molliq.2025.128934.
(https://www.sciencedirect.com/science/article/pii/S0167732225021117)
Abstract: Sinomenine hydrochloride (SH) is an alkaloid derived from the traditional Chinese medicinal plant Sinomenium acutum. Clinically, SH is widely used to treat chronic conditions such as rheumatism and bone pain. However, its strong water solubility poses a significant challenge for its direct application in sustained-release formulations. Additionally, while SH has been explored for use in transdermal patches, its high crystallinity complicates the preparation process, limiting its practical application. Deep eutectic solvents (DESs) have emerged as a promising, eco-friendly intermediate in pharmaceutical formulations due to their natural and green characteristics. Nevertheless, the screening of suitable DES ligands remains a time-consuming and labor-intensive process. To address this issue, we propose the development of a machine learning-based model for the rapid screening of DES ligands, with the ultimate goal of formulating a DES-based SH delivery system. In this study, we employed six widely used machine learning algorithms to construct a predictive model for DES ligand screening. Using this model, we efficiently screened 11 commonly used DES ligands and identified citric acid as a suitable candidate for forming a DES with SH. The model's prediction was subsequently validated by experimental data, confirming the successful formation of a DES between SH and citric acid. Our findings demonstrate that the machine learning model, based on molecular descriptors, is capable of rapidly and accurately screening DES ligands, offering a valuable tool for optimizing pharmaceutical formulations.
Keywords: Sinomenine hydrochloride; Deep eutectic solvent; Machine learning; Ligand screening; Molecular descriptor