用于药物纳米颗粒生产的小分子活性药物成分在超临界溶剂中溶解度的计算智能建模与优化

Computational intelligence modeling and optimization of small molecule API solubility in supercritical solvent for production of drug nanoparticles.

作者信息

Liang Jiaxin, Zhao Hai, Zhou Shengxue, Gao Yang

机构信息

College of Chinese Medicine, Jilin Agricultural Science and Technology College, Jilin, 132101, China.

Department of Burn and Plastic Surgery, The General Hospital of Northern Theather, Shenyang, Liaoning, 110100, China.

出版信息

Sci Rep. 2025 Apr 28;15(1):14774. doi: 10.1038/s41598-025-99776-1.

DOI:10.1038/s41598-025-99776-1

PMID:40295721

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC12037993/

Abstract

Artificial Intelligence (AI) is applied in this research for the analysis of a novel green method for production of nanomedicine. The method is based on supercritical solvent for production of drug nanoparticles in which the AI was used to estimate the solubility of drug in the supercritical solvent. Carbon dioxide was considered as the supercritical solvent in this study and the effect of pressure and temperature on the drug solubility was evaluated using the developed AI-based models. The aim here is to model and analyze the solubility of Clobetasol Propionate (CP) based on two key parameters: temperature and pressure. Ensemble models based on decision tree are selected to make models. The models include gradient boosting (GBDT), extremely randomized trees (ET), and random forest (RF) and tuned using ant colony optimization (ACO). Final models have acceptable results, all with R criterion more than 0.9. The GBDT model outperforms the others with an R of 0.987. Additionally, the RMSE for this model is minimized to 8.21 × 10.

摘要

本研究应用人工智能（AI）分析一种新型纳米药物绿色生产方法。该方法基于超临界溶剂制备药物纳米颗粒，其中利用人工智能估计药物在超临界溶剂中的溶解度。本研究将二氧化碳视为超临界溶剂，并使用基于人工智能开发的模型评估压力和温度对药物溶解度的影响。此处的目的是基于温度和压力这两个关键参数对丙酸氯倍他索（CP）的溶解度进行建模和分析。选择基于决策树的集成模型来构建模型。这些模型包括梯度提升决策树（GBDT）、极端随机树（ET）和随机森林（RF），并使用蚁群优化（ACO）进行调优。最终模型具有可接受的结果，所有模型的R准则均大于0.9。GBDT模型表现优于其他模型，R值为0.987。此外，该模型的均方根误差（RMSE）最小化至8.21×10。