Sustainable valorisation of agricultural residues into polyhydroxybutyrate and amylase using XGBoost-optimised extractive co-fermentation and its macromolecular characterization.

As food production increases globally, the disposal of agro-waste is an additional issue. However, turning this waste into products through sustainable methods could probably reduce environmental impact. In this work, the inexpensive carbon sources rice bran and corn bran were hydrolysed to produce Polyhydroxybutyrate (PHB) and Enzyme (Amylase) by extractive Co-fermentation. This work is advanced by the synthesis, extraction, and characterization of PHB produced by the strain Priestia megaterium isolated from soil. A 48 h incubation in modified medium, optimized PHB and enzyme production, achieving a yield of up to 83.10% and 81.37% respectively. The response surface methodology (R2 = 0.96) and machine learning models for enzyme recovery, incorporating PEG concentration, salt concentration, and temperature (30 °C), yield a prediction process with 95% accuracy using XGBoost (R2 = 0.9501).The enzyme yield was modeled through PEG and salt concentrations, with the temperature at 30 °C, underwent prediction studies by RSM (R2 = 0.9600), and machine learning (R2 = 0.9501) models. The PHB recovered by methanol precipitation was characterized by FTIR, GC-MS, HPLC, NMR, and GPC. The recovered enzyme was purified by gel filtration chromatography, yielding an 83.45% recovery. This study confirmed that the PEG-based Extractive Co-fermentation with improved efficiency and affordability for PHB and enzyme production. The recycling of agro-waste into valuable products supports the circular bio economy by maximizing resource efficiency and directly aligns with SDG 12 and SDG 13 through sustainable production and climate-action strategies.