Artificial Neural Network- Response Surface Methodology based multi-parametric optimization and modelling of biolipid production from Aspergillus flavus

Abstract

Microbial lipids, produced by oleaginous microorganisms, are emerging as sustainable feedstocks for biodiesel and other industrial applications. In this study, biolipid production from Aspergillus flavus was systematically optimized through cultivation studies, lipid extraction, and quantification, combined with classical and advanced modeling approaches. Key nutrients such as carbon sources, nitrogen sources, amino acids and metal salts were analyzed for their influence on lipid production. Optimization studies were performed using Response Surface Methodology (RSM) and Artificial Neural Network (ANN) models. RSM, employing Plackett-Burman design and Central Composite Design (CCD), identified critical parameters (pH, glucose, and peptone) affecting lipid yield, achieving high predictive accuracy with an R2 value of 0.9911. The ANN model, with a configuration of 17 hidden neurons, outperformed RSM, yielding correlation coefficients (r) of 0.999 for training and validation and 0.981 for testing, along with lower Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). Further, 3D contour plots and sensitivity analysis elucidated the interactive and non-linear effects of key parameters. This integrated approach demonstrates the superiority of combining statistical (RSM) and computational (ANN) tools for bioprocess optimization. The study highlights A. flavus as a promising microbial resource for sustainable lipid production, providing a scalable framework for industrial biodiesel manufacturing.