AI-Driven Virtual Depository of Medicinal Herbs for AYUSH

Abstract

Medicinal plants are a cornerstone of the AYUSH healthcare systems—Ayurveda, Yoga & Naturopathy, Unani, Siddha and Homeopathy—serving as vital resources for both preventive and therapeutic practices. Despite their significance, existing knowledge about these herbs is dispersed across classical manuscripts, botanical collections, traditional community practices and various digital archives. This fragmentation creates barriers to reliable identification, standardization, and safe clinical application. To address these issues, this study introduces HerboAI, an AI-powered Virtual Repository for Medicinal Plants. The platform is designed as a centralized, multilingual, and scalable system that organizes, classifies, and retrieves structured information on herbal resources. Herbo AI integrates multiple AI technologies: computer vision enables species recognition from plant images, natural language processing (NLP) extracts useful knowledge from unstructured documents and ontology-based knowledge graphs unify information from diverse sources into a coherent framework. The system architecture follows a modular three-layer design: a user-friendly web interface for interaction, an intelligent backend that supports semantic search and retrieval and a structured database aligned with AYUSH guidelines. The platform allows both professionals and general users to submit natural language queries, providing context-aware, verified, and curated recommendations. Initial experiments demonstrate that HerboAI’s offline AI modules perform effectively in semantic search, species classification, and recommendation tasks, achieving strong accuracy and user satisfaction. Beyond immediate utility, the platform holds broader promise for accelerating drug discovery, conserving rare or endangered plant species, reducing herbal adulteration, and safeguarding traditional medical knowledge, all while ensuring ethical and legal compliance.