Integrated machine learning models for optimizing and assessing microbial remediation of indigo dye

The textile industry produces waste contaminated with toxic dye chemicals like indigo which presents an environmental challenge because of their persistence and resistance. However, current physiochemical degradation techniques are not only energy-intensive but may generate other forms of hazardous secondary wastes, therefore there is need to adopt sustainable remediation techniques through bioremediation. This study presents investigation of aerobic biodegradation of indigo dye by a unique bacterial consortium (20B) with the aid of data-driven predictive models for optimizing the process conditions. Data used in this work involved 50 experiments on different environmental and nutritional factors considered, such as pH, temperature, inoculums' concentrations, glucose, and NH₄Cl. Boosted Regression Tree (BRT) algorithm showed better prediction than other models tested giving R²=0.95. Therefore, under optimized conditions of pH = 6.48, temperature of 36.25 °C, inoculums of 14.98%, glucose concentration of 23.79 g/L, and NH₄Cl of 15.23 g/L, the consortium attained 100% decolurization of indigo dye at concentration of 200 mg/L after 12 h incubation under aerobic condition. Wastewater treatment efficiency was significant since Chemical Oxidation Demand (COD) decreased by 93-97% and Biological Oxygen Demand (BOD) was removed by 96-99%.