Employing RSM and ANN-based applications for modelling enhanced bacterial cellulose production from pineapple peel waste using Komagateibacter saccharivorans APPK1

Abstract

Bacterial cellulose (BC) is a bio-polymer with variety of applications in diverse fields such as biomedical, cosmetics, environment and food. Owing to its superior crystallinity, purity and other unique properties BC outshines plant cellulose in several aspects. The current study focuses on a cost-effective method for BC production from pineapple waste using Komagateibacter saccharivorans APPK1. The research includes a systematic optimisation approach that includes screening of various components in the medium. The variables under investigation in this study were ammonium sulphate, ethanol volume and incubation days. These variables were selected and evaluated using the Plackett-Burman experimental design model. Additionally, the engineering of process parameters using a process-driven approach i.e., response surface methodology (RSM) and a data-driven approach i.e., artificial neural network (ANN) were employed. BC production was increased up to 43 % with a maximum yield of 47.5 g/100 mL using RSM. However, on comparing RSM and ANN models for enhanced BC production based on error functions such as R2 value, mean absolute percentage error, root mean square error, and mean absolute deviation values, the ANN model emerged as the superior one. As this is the first report to the authors’ best knowledge on model development using a hybrid approach of RSM and ANN for enhanced BC production using Komagateibacter saccharivorans APPK1, this could lead to more in-depth and large-scale production of BC.