Sindhu Parupalli, MD1, Samar A. Hegazy, MD, PhD2 1University of Illinois, Chicago, IL; 2Carle Illinois College of Medicine, Urbana, IL
Introduction: Epstein-Barr virus-associated gastric cancer (EBVaGC) exhibits unique molecular and histopathological features which can be identified to predict response to immunotherapy. Universal testing for EBVaGC is expensive and labor-intensive. Artificial intelligence (AI) is used in pathology explorations, including diagnostic validation and prediction of treatment response. Recently, AI applications have been used to better understand and manage EBVaGC. This study undertakes a scoping review of research on AI applications in pathology explorations of EBVaGC to determine how AI is currently being used to contribute to improved diagnosis and targeted therapeutic interventions in patients with EBVaGC.
Methods: The PRISMA method for scoping reviews was selected. The PubMed database was searched to identify publications that address this topic considering the following inclusion criteria: full-text availability, English language, in any timeframe of publication, any type of AI exploration, EBVaGC. This strategy resulted in ten total publications due to the recency of AI innovation in pathology, and all ten publications were screened and reviewed.
Results: Half of the publications (50%) reported diagnostic validation studies, 20% reported retrospective multicenter studies, 20% reported comparative analysis studies, and 10% reported observational studies. All studies involved data analysis. AI applications in EBVaGC include a machine learning approach with algorithms, swarm learning and a deep learning approach with convolutional neural networks, transformers, classifiers, and segmentation. The main clinical applications include diagnostic studies for pre-screening patients for confirmatory EBV testing, precision medicine with biomarkers to predict therapeutic response, and molecular characterization.
Discussion: AI is a powerful tool that can reduce labor and cost associated with pathology testing for EBVaGC. The utility of AI in EBVaGC includes triaging patients with confirmatory testing, identifying biomarkers to predict response to immunotherapy, and providing molecular characterization. AI advancements are unlikely to replace pathologists or gastroenterologists; however, they have the potential to augment pathology explorations and personalized therapies in EBVaGC by increasing efficiency and decreasing burnout.
Disclosures:
Sindhu Parupalli indicated no relevant financial relationships.
Samar Hegazy indicated no relevant financial relationships.
Sindhu Parupalli, MD1, Samar A. Hegazy, MD, PhD2. P5047 - Artificial Intelligence Applications in Epstein-Barr Virus-Associated Gastric Cancer: A Scoping Review, ACG 2024 Annual Scientific Meeting Abstracts. Philadelphia, PA: American College of Gastroenterology.
