Comparison of Laboratory-Acquired Zif-8 Drug Delivery-Related Parameters With Literature-Based Data Via Machine Learning Tools Publisher



N Pouyanfar NIKI ; N Masoumi NILOOFAR ; M Ahmadi MAHNAZ ; Sm Ayyoubzadeh Seyed MOHAMMAD ; G Farnam GOLROKH ; E Asadian ELHAM ; F Sefat FARSHID ; Sf Hosseini Shirazi Seyed FARSHAD ; F Ghorbanibidkorpeh FATEMEH
Source: Materials Today Chemistry Published:2025

Abstract

Nanotechnology has been moving forward at a fast pace in recent years. Advancements in this field have led to the development of various nanostructures in diverse fields. Nanomedicine, as an important area of research, has resulted in great progression in disease treatment, from cancer to tissue regeneration. Coupling this technology with artificial intelligence (AI) has resulted in major steps that could ease the process of synthesis, characterization, and in vitro and in vivo evaluation of the formed nanoparticle, thus reaching better outcomes with less effort and cost. Metal-organic frameworks (MOFs) are among the nanostructures with important features ranging from great loading and porosity to imaging capabilities and modification potential. Facilitating the synthesis of MOF nanoparticles with novel techniques and AI assistance may have a great impact on the future of nanomedicine in general. In this study, MOF nanoparticles prepared via two different routes (room temperature and microfluidics) and loaded with different drugs (metformin hydrochloride and rosuvastatin calcium) were characterized and assessed in vivo, the results were then compared to the predictions made with machine learning methods, to open up ways for utilization of AI tools without laboratory efforts. Using machine learning techniques such as Random Forest and XGBoost, we achieved improved prediction accuracy for encapsulation efficiency (RMSE = 14.72) and cumulative drug release (RMSE = 18.91). SHAP analysis highlighted critical parameters such as drug hydrophobicity, particle size, and synthesis route as major influencers. © 2025 Elsevier B.V., All rights reserved.