Abstract Gastric cancer (GC) remains a leading cause of morbidity and mortality within the global digestive system. Early screening is critical to improve patient prognosis and reduce mortality. In recent years, advances in big data and artificial intelligence have underpinned the creation and application of innovative scoring systems for GC screening. These tools are designed to enable accurate risk stratification, highlighting their specific value in facilitating early detection among low-risk groups. However, such systems still carry a non-negligible risk of missed diagnoses, delaying detection and impacting screening efficacy and safety. This review systematically examines the foundational principles and current applications of newly developed scoring systems for GC screening. It further delves into mechanisms contributing to missed diagnoses in low-risk groups, including limitations in model design – such as feature selection and threshold setting – and the complexity of tumor biological behavior, such as heterogeneity and progression variability in low-risk GCs. Additional factors, including insufficient clinical samples and data, as well as constrained sensitivity and specificity of current screening technologies, are explored in depth. By synthesizing recent literature and research advances, this review identifies the principal mechanisms behind screening omissions in low-risk individuals. It also provides theoretical foundations and clinical recommendations aimed at optimizing screening tools and refining early detection strategies, thereby promoting precision and personalization in GC screening.
Shao et al. (Thu,) studied this question.