Optimization of starch extraction from Amorphophallus paeoniifolius corms using response surface methodology (RSM) and artificial neural network (ANN) for improving yield with tenable chemical attributes.

The development of the extraction process for improving the starch yield from unconventional plants is emerging as a topic of interest. In this respect, the present work aimed to optimize the starch extraction from the corms of elephant foot yam (Amorphophallus paeoniifolius) with the help of response surface methodology (RSM) and artificial neural network (ANN). The RSM model performed better than the ANN in predicting the starch yield with higher precision. In this connection, this study for the first time reports the significant improvement of starch yield from A. paeoniifolius (51.76 g/100 g of the corm dry weight). The extracted starch samples based on yield - high (APHS), medium (APMS), and low (APLS) exhibited a variable granule size (7.17-14.14 μm) along with low ash content, moisture content, protein, and free amino acid indicating purity and desirability. The FTIR analysis also confirmed the chemical composition and purity of the starch samples. Moreover, the XRD analysis showed the prevalence of C-type starch (2θ = 14.303°). Based on other physicochemical, biochemical, functional, and pasting properties, the three starch samples showed more or less similar characteristics thereby indicating the sustentation of beneficial attributes of starch molecules irrespective of the variation in extraction parameters.