AI-based strategies for diagnosis and therapeutic discovery in pulmonary fibrosis

Pulmonary fibrosis (PF) is a life-threatening and therapeutically challenging disease, underscoring the urgent need for novel and effective interventions. Natural products, characterized by structural diversity, wide biological activities, and abundant availability, represent a promising reservoir for anti-fibrotic drug discovery. This review provides a comprehensive overview of the key risk factors and complex pathogenic mechanisms underlying PF, with a particular focus on the pro-fibrotic and anti-fibrotic targets that regulate disease progression. Special attention is given to high-value candidate targets identified through data-driven approaches, including members of the G protein-coupled receptor (GPCR) family. This study further summarizes the emerging roles of artificial intelligence (AI) in PF diagnosis and prognostic assessment, where deep learning models significantly enhance imaging interpretation and patient stratification. In addition, the anti-fibrotic mechanisms of natural products are systematically discussed, especially their modulation of fibroblast-to-myofibroblast transition (FMT), epithelial-mesenchymal transition (EMT), and endothelial-to-mesenchymal transition (EndMT). Building on these insights, this study proposes an AI-driven workflow for natural product-based anti-fibrotic drug discovery, integrating target identification, virtual screening, and experimental validation. This integrative strategy offers a conceptual and technical framework for accelerating the development of natural-product-derived therapeutics for PF.