Ultrasound-Assisted Extraction of Bioactives from Spirulina platensis: Optimization and Prediction of Their Properties Using Near-Infrared Spectroscopy Coupled with Artificial Neural Network Modeling

This study optimizes the ultrasound-assisted extraction (UAE) of bioactive compounds from <i>Spirulina platensis</i> and develops a rapid, non-destructive analytical method. A Box-Behnken design and desirability function were used to find the optimal extraction conditions to simultaneously maximize total polyphenols, proteins, C-phycocyanin, and antioxidant activity. The optimal conditions were a solid-to-liquid ratio of 35 g/L, a time of 20 min, a pH of 10, and a temperature of 45 °C. Independent validation confirmed the model's reliability, with experimental results closely matching predicted values. Furthermore, Near-Infrared (NIR) spectroscopy, combined with Artificial Neural Networks (ANNs), was explored as a predictive tool. The models, particularly those based on a semi-process NIR spectrometer, showed excellent predictive capabilities for key physicochemical properties, with an RPD of 3.9064 for <i>L</i>* and 2.8351 for TDS. This research establishes a highly reproducible and scalable extraction protocol, complemented by a fast and accurate analytical method, providing a significant advancement for the industrial application and quality control of <i>Spirulina platensis</i> extracts.