Enhancing the functional and structural properties of arrowroot starch through DBD cold plasma treatment: an ANN–PSO modeling approach

Abstract

Arrowroot starch possesses considerable potential for food, pharmaceutical, and biodegradable packaging applications; however, its industrial utilization is constrained by the limited functionality of native starch. Although cold plasma treatment has been reported to improve the physicochemical properties of arrowroot starch, studies integrating process optimization with intelligent predictive modeling remain scarce. Therefore, this study aimed to enhance the functional properties of arrowroot starch through dielectric barrier discharge (DBD) cold plasma treatment and optimize the process conditions using a hybrid Artificial Neural Network–Particle Swarm Optimization (ANN–PSO) approach. The effects of treatment on physicochemical, thermal, structural, and techno-functional properties were also evaluated. Arrowroot starch was treated at different voltages of DBD (15–25 kV), exposure times on DBD (2–10 min), and temperatures of water bath (70–90 °C). An ANN model was developed to predict swelling power and solubility, while PSO was employed for multi-objective optimization. Cold plasma treatment significantly modified starch characteristics by increasing crystallinity, water and oil uptake capacities and thermal stability while reducing pH and turbidity. The optimized ANN architecture (3-16-8-2) demonstrated excellent predictive performance ranging from R 2 of 0.93–0.98 across the training and testing datasets. The R 2 score for training and testing datasets of solubility and swelling power were 0.966, 0.926, 0.982 and 0.960, respectively. Multi-objective ANN–PSO optimization identified optimum processing conditions that maximized either solubility (49.94%) at 19.99 kV, 2 min and 90 °C or swelling power (33.05 g/g) at 15 kV, 10 min and 90 °C, with experimental validation showing prediction errors below 2%. The findings demonstrate that DBD cold plasma is an effective green modification technology for tailoring arrowroot starch functionality, while the ANN–PSO framework provides a robust tool for process optimization. This integrated approach offers a sustainable pathway for developing customized starch ingredients for advanced food and industrial applications.