# Microencapsulation of Beetroot Anthocyanins: Investigation of Degradation Kinetics and Modeling by Using Artificial Neural Networks

## TL;DR

This study explores how to stabilize beetroot anthocyanins using microencapsulation and finds that a specific formulation improves stability and predicts degradation using an AI model.

## Contribution

A novel ternary microencapsulation system (MD/GA/SC) and an AI model for predicting anthocyanin degradation are introduced.

## Key findings

- The MD/GA/SC formulation achieved the highest encapsulation efficiency and antioxidant activity.
- The ANN model outperformed traditional models in predicting anthocyanin degradation with high accuracy.
- The MD/GA/SC system showed significantly improved thermal stability across various pH levels.

## Abstract

Anthocyanins are
widely appreciated as natural pigments, but their
use in foods and related industries is still quite limited because
they are highly sensitive to heat, pH changes, light, and oxygen.
Improving their stability has therefore become a key focus in developing
more reliable natural color systems. In this study, beetroot anthocyanins
were microencapsulated with different wall materials, maltodextrin
(MD), gum arabic (GA), a simple MD/GA blend, and a ternary structure
combining MD, GA, and sodium caseinate (MD/GA/SC). These systems were
evaluated for their encapsulation efficiencies, antioxidant activity
preservation, release behaviors, and degradation responses over a
wide range of temperatures (40–100 °C) and pH levels (2.5–6.5).
Remarkable findings demonstrated that the MD/GA/SC formulation provided
the highest encapsulation efficiency (93.36%), superior radical-scavenging
activity (88.43%), and the most controlled release profile. Moreover,
this formulation demonstrated the lowest degradation rate constants
at pH 2.5, 4.5, and 6.5 (2.886, 2.083, and 1.30 1/min, respectively)
together with the highest activation energies at these pH levels (37.460,
52.517, and 62.045 kJ/mol, respectively), indicating a pronounced
improvement in thermal stability compared with the other formulations
and the free extract. An artificial neural network (ANN) model was
developed to predict anthocyanin degradation. The ANN provided highly
accurate predictions (R
2 > 0.98, RMSE
< 0.01) across all conditions and outperformed the classical first-order
kinetic model. These findings highlight the potential of the MD/GA/SC
matrix as a promising encapsulation system for improving anthocyanin
stability. The strong performance of the ANN model also suggests that
data-driven approaches can contribute meaningfully to designing more
reliable microencapsulation strategies for future food and nutraceutical
applications.