Development of multilayer micro- and nanocapsule structures and evaluation of permeation and release behavior based on experimental data analysis and advanced simulation with convolutional neural networks (CNN) for dietary supplements and functional beverages

behzad beizaei¹ morteza jamshid eini²

1) Master s student in Food Industry, Islamic Azad University, Science and Research Branch
2) PhD student in Food Technology, Islamic Azad University, North Tehran Branch

IOI Code : AADK-AAFAHE
Publication : 3rd International Congress on Science, engineering & New Technologies(secongress.com/3rd)
Abstract :
    This research deeply focuses on the development and optimization of hybrid multilayer micro- and nanocapsule structures capable of precise and targeted control over the kinetics of diffusion and release of bioactive compounds, which are highly sensitive to physicochemical environmental factors such as temperature, relative humidity, pH, and electrochemical potential. In this context, advanced deep learning frameworks, especially multi-layer convolutional neural networks (CNNs), have been used to model experimental data with high temporal and spatial resolution, in order to quantitatively and qualitatively extract the nanomorphological patterns influencing molecular diffusion within each capsule layer. The training process of the neural network has been designed by employing optimized hyperparameter tuning techniques, including adaptive learning rates, overfitting reduction capabilities via dropout mechanisms, and intelligent layer weighting, ensuring high prediction accuracy of the dynamic release behavior of bioactive compounds. Computational results demonstrate a significant statistical match between the modeled and experimental data, validated with fundamental error indices such as root mean square error (RMSE) and multi-criteria sensitivity analysis of process parameters, enabling management of the spatiotemporal distribution of release rates and probabilistic distribution of active compounds within each layer. This interdisciplinary approach, integrating nanomaterials technology, polymer chemistry, and machine learning, provides the capacity to design interactive, customizable nanostructures to meet requirements for autopharmacological properties, biological stability, and environmental reliability. Moreover, the presented technology, while increasing the bioactive compounds lifespan and reducing production costs, creates an innovative platform for the new generation of smart functional food industries with precise control capabilities in bioactive material delivery systems, considered a milestone in advanced food engineering.

Keywords : multilayer nanoencapsulation convolutional neural network diffusion kinematics controlled release biodegradable polymers dietary supplements functional beverages deep learning modeling