– published in Journal of Ethnopharmacology
Summary
Can artificial intelligence actually decode the molecular secrets behind ancient Ayurvedic concepts like herbal potency (Vīrya)?
SitoshnaPred, a machine learning framework, successfully predicts Ayurvedic potencies Sīta (cold) and Uṣṇa (hot) based on phytochemical constituents. Assuming traditional potency reflects molecular properties, researchers classified 627 herbs and trained a LightGBM classifier using molecular descriptors. The model achieved high accuracy (94.01%), confirming a strong correlation between an herb’s chemical makeup and its classical Vīrya classification. This work validates the molecular basis of ancient Ayurvedic wisdom using AI.