Predictive optimization of curcumin nanocomposites using hybrid machine learning and physics informed modeling.
Saved in:
| Title: | Predictive optimization of curcumin nanocomposites using hybrid machine learning and physics informed modeling. |
|---|---|
| Authors: | Rahdar, Abbas, Fathi-Karkan, Sonia, Shirzad, Maryam |
| Source: | Scientific Reports; 12/23/2025, Vol. 15 Issue 1, p1-13, 13p |
| Subject Terms: | CURCUMIN, MACHINE learning, PREDICTION models, NANOCOMPOSITE materials, EXPERIMENTAL design, COMPUTER simulation, BOOSTING algorithms, MATHEMATICAL optimization |
| Abstract: | The aim of the current research was to develop a hybrid computational model that involves the integration of Machine Learning (ML) with Physics-Informed Neural Networks (PINNs) to predict and maximize curcumin nanocomposite performance based on Loading Efficiency (LE%) and Encapsulation Efficiency (EE%). Quantitative experimental design comprising 74 synthesized nanocomposite formulations, selected by power analysis (α = 0.05, power = 0.8) to allow for statistical reliability. Pre-processing of data and model construction was undertaken in Python (v3.11) with libraries such as scikit-learn, TensorFlow, and SHAP (as downloaded from the Python Software Foundation). Different ML regressors were tested, and the highest predictive power was manifested by the Gradient Boosting Regressor (GBR). A custom-defined PINN was constructed to integrate mechanistic understanding from the Derjaguin–Landau–Verwey–Overbeek (DLVO) theory and diffusion-transport constraints. The hybrid model was also elucidated by SHapley Additive exPlanations (SHAP) analysis to find controlling formulation parameters. The optimized integrated model demonstrated extremely good predictive performance (LE%: R2 = 0.89, RMSE = 6.24; EE%: R2 = 0.87, RMSE = 7.15). Physical constraints enhanced generalization by 23%, confirming robustness of the model. Significant optimization results revealed optimal design parameters with particle diameters ranging from 80 to 200 nm and zeta potentials ranging from − 30 to − 50 mV. The data mining analysis revealed polymer ratio and surfactant concentration to be the most influential variables, consistent with the predictive equation derived in the full text. This hybrid ML–PINN combines the predictive capability of machine learning with the mechanistic interpretability of physics-informed modeling. This is a stable, comprehensible optimization platform for nanocarrier optimization with 40–60% cost reduction in experimental screening. Further research is recommended to validate the framework to other bioactive drugs and for incorporating real-time adaptive optimization for scalable pharmaceutical manufacture. [ABSTRACT FROM AUTHOR] |
| Copyright of Scientific Reports is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Complementary Index |
Full Text 
Full Text Finder 
Nájsť tento článok vo Web of Science Be the first to leave a comment!