Sustainable hydroxychloroquine removal by living Chlorella sp. microalgae: Experimental investigation, box– B ehnken optimization, and machine learning

Abstract

Pharmaceutical pollutants like hydroxychloroquine (HCQ) pose increasing environmental risks due to their persistence in aquatic systems. This study integrates Box–Behnken design (BBD) and machine learning (ML) to optimize and predict HCQ removal from water using living Chlorella sp. microalgae. Three key variables, initial HCQ concentration, pH, and microalgae dosage, were investigated for their individual and interactive effects on removal efficiency. BBD results identified pH as the dominant factor, with maximum removal (>90%) achieved at pH 11. The quadratic model was highly significant ( R ²  = 0.998) and predicted an optimal removal of 92.06% at pH 11, 5 mg L ⁻¹ HCQ, and 75 mg L ⁻¹ microalgae. Validation experiments using real wastewater spiked with 5 mg L ⁻¹ HCQ reached 89.76% removal, confirming the model's robustness under realistic conditions. Concurrently, five ML algorithms, Linear Regression, Support Vector Regression, Random Forest (RF), Gradient Boosting Regression, and XGBoost, were developed and evaluated on an expanded dataset. All models performed well ( R ²  >0.92), with RF achieving the highest accuracy ( R ²  = 0.987, RMSE = 3.63, MAE = 2.24), effectively capturing the nonlinear relationships between operating parameters and HCQ removal. The removal mechanism involves biosorption, bioaccumulation, and biodegradation. Overall, the synergy between response surface methodology and machine learning offers a reliable framework for optimizing and predicting microalgae‐based HCQ removal, providing a promising approach for treating pharmaceutical‐contaminated wastewater.