Predicting and Optimizing Zerumbone Content in Zingiber zerumbet Using Artificial Neural Network (ANN) Model: A Statistical Approach and Experimental Validation.

The pharmacological potential of Zingiber zerumbet is primarily attributed to its key bioactive compound, zerumbone. Zerumbone is known for its antioxidant, antiproliferative, anti-inflammatory, antimicrobial, and anticancer properties. The zerumbone content varies significantly across habitats. This study employed an artificial neural network (ANN) to assess the influence of abiotic factors, such as environmental parameters and soil nutrients, on zerumbone content and to predict optimal cultivation sites for Z. zerumbet. Sixty rhizome accessions from diverse locations in Eastern India were analyzed using 18 input variables, including soil nutrients and climatic parameters. The dataset was randomly divided into 70, 15, and 15% for training, testing and validation, respectively. HPTLC analysis revealed a zerumbone content ranging from 0.91 ± 0.04% to 21.46 ± 0.27% (dry weight). A multilayer perceptron model with an 18-11-1 architecture achieved high predictive accuracy (R2 = 0.978) and 96% efficiency. Sensitivity analysis identified manganese (Mn), zinc (Zn), and iron (Fe) as key influencers. Optimizing these factors within the model increased predicted zerumbone content from 16.10% to 18.51%. The ANN model offers a robust tool for site-specific cultivation strategies to maximize zerumbone yield in Z. zerumbet.