[Current status and challenges of artificial intelligence and organoid technologies in precision diagnosis and treatment of gastrointestinal stromal tumors].

Gastrointestinal stromal tumor (GIST) is the most common mesenchymal tumor of the gastrointestinal tract. Current diagnostic and therapeutic approaches mainly rely on surgical resection and targeted therapy. However, early clinical diagnosis and resistance to targeted drugs remain the most challenging issues, severely compromising patient outcomes. Organoid technology can preserve patient-specific genotypic characteristics and drug sensitivity profiles in vitro, providing a novel platform for drug screening and investigation of resistance mechanisms; however, its application in GIST research remains at an early stage. Artificial intelligence (AI)-based radiomics, pathomics, large language models, and intelligent agents have demonstrated substantial potential in GIST risk stratification, genotype prediction, treatment response evaluation, and whole-lifecycle disease management. Integrating organoid model-derived data with multimodal clinical data through AI approaches and advancing the development of personalized decision-support systems represent emerging directions in precision diagnosis and treatment of GIST. The establishment of GIST organoid models is currently limited to case reports, with low culture success rates and prolonged drug sensitivity testing periods, making it difficult to meet the requirements of timely clinical decision-making. In contrast, AI applications in GIST imaging-based risk stratification and pathology-based genotype prediction have demonstrated preliminary high performance; however, most models still lack multicenter external validation, and their generalizability remains uncertain. Currently, the integration of AI and organoid technologies, including morphological quantitative assessment and virtual drug screening, relies mainly on methodological approaches developed in pan-cancer studies, with no studies directly investigating these approaches in GIST. Moreover, the development of multimodal decision-support systems faces major challenges, including data silos, insufficient algorithm interpretability, and the absence of clear regulatory pathways. Therefore, future efforts should focus on establishing standardized databases through multicenter collaboration and validating the clinical utility of integrated platforms through real-world studies, thereby facilitating the transition of these technologies from conceptual frameworks to clinical practice.