Surfactant-facilitated metabolic induction enhances lipase production from an optimally formulated waste-derived substrate mix using Aspergillus niger: A case of machine learning modeling and metaheuristic optimization

Abstract

This study introduces a novel method to enhance lipase production by integrating machine learning (ML) models and optimization algorithms. Six ML models, including support vector regression (SVR), kernel ridge regression (KRR), extreme gradient boosting (XGB), extreme learning machine (ELM), random forest (RF), and artificial neural networks (ANN), were employed to predict lipase activity in a multi-substrate system using avocado seed, coconut pulp, and palm oil mill effluent (POME) with Aspergillus niger. SVR proved the most effective (R² = 0.9738; RMSE = 7.0089). Further optimization using manta ray foraging optimization (MFRO), particle swarm optimization (PSO), and genetic algorithm (GA) identified optimal substrate loadings, achieving a maximum lipase activity of 194.38 U/g_ds. The addition of a mixture of surfactants (Tween 80, Tween 20, Triton X-100) further increased lipase production to 520.95 U/g_ds (168.3 % increase). Global sensitivity analysis (GSA) confirmed the important roles of avocado seed, POME, and surfactants in enhancing lipase production. This approach represents a significant advancement in bioprocess scalability.