Biotechnological Evaluation and Predictive Modeling of Bio-Based Adsorbents for Turquoise Blue Dye Detoxification: Integrating Experimental Validation and Machine Learning

This study presents a data-driven hybrid framework that combines experimental adsorption trials with machine learning and evolutionary optimization to enhance the removal of Turquoise Blue Dye from aqueous media. Three bio-adsorbents, Azadirachta indica, Phyllanthus emblica, and Saraca asoca, and their equi-mass mixture were examined for synergistic adsorption performance. Batch experiments were conducted by varying the pH (3-11), contact time (20-120 minutes), dye concentration (20-120 ppm), and adsorbent dosage (0.02-0.1 g/L). The composite mixture achieved the highest removal efficiency at 77% exceeding Phyllanthus emblica at 74%, Saraca asoca at 70%, and Azadirachta indica at 63.8%. FTIR analysis confirmed chemical interactions via hydroxyl, carboxyl and carbonyl groups. Supervised models, including Random Forest with a coefficient of determination of 0.92 and mean squared error of 0.0021, were optimized using Differential Evolution. This integrative strategy supports scalable and intelligent solutions for industrial dye effluent remediation.