Segmented microwave hot-air rolling drying of <i>Gastrodia elata</i> : Effects on quality, water migration, and real-time moisture prediction via LSTM network

This study investigated segmented drying strategies for Gastrodia elata (RGE), combining microwave hot-air rolling-bed drying (MHRD) with hot-air drying (HD). Four transition points (60%, 45%, 30%, and 15% moisture content) were applied to examine their effects on drying performance, moisture migration, rehydration, and bioactive retention. The segmented approach significantly altered drying kinetics and water state dynamics. Under the drying conditions of microwave power density 0.8 W/g and hot-air velocity 2.5 m/s, the mid-to-late transition points, particularly at 15%, produced dried products with enhanced color (L*=51.98), rehydration ratio (4.84), and microstructure. Chemical profiling confirmed its high contents of phenolic (3.03 g GAE/kg), gastrodin (4.422 g/kg), and p-hydroxybenzyl alcohol (1.413 g/kg). An LSTM model was developed to predict moisture content and moisture ratio. The model achieved high predictive accuracy (R2 > 0.99). This work provides insight into how MHRD-HD influences the dried RGE, offering a practical framework for intelligent drying through AI-assisted moisture prediction.