Application of artificial neural network (ANN) model for prediction and optimization of coronarin D content in Hedychium coronarium

The pharmacological properties of Hedychium coronarium Koen. is due to the presence of its active constituent Coronarin D. Coronarin D has been found to possess a myriad of therapeutic activities ranging from antimicrobial to anticancer. Coronarin D content in H. coronarium greatly differs in different habitat. In this study, an artificial neural network (ANN) based model was developed to investigate the influence of abiotic factors (climate and soil) and predict a suitable region for cultivation of H. coronarium with high content of coronarin D. The experimental dataset of 50 was generated by collecting H. coronarium rhizomes from 50 different geographical locations distributed in five different states of India. For each location, 18 input parameters were considered including soil nutrients (micronutrients and macronutrients) and climatic factors. Datasets were randomly partitioned with 72 %, 14 % and 14 % for training, validation and testing dataset, respectively. HPTLC analysis revealed coronarin D content to vary from 0.136 to 0.687 mg/100 mg dry wt among 50 H. coronarium rhizomes. Results showed that the multilayer perception (MLP) neural network with single hidden layer containing 5 neurons namely 18-5-1 structure could predict the coronarin D content accurately with a correlation coefficient (R-2) of 0.891 and root mean square error (RMSE) of 0.06. Sensitivity analysis revealed the effect of altitude, manganese and zinc on predicted coronarin D content to be slightly higher compared to other factors. The developed ANN model will assume a great significance in the prediction of the proper regions/site for optimum coronarin D yield in H. coronarium.