Advances and Challenges in Artificial Intelligence and Statistical Methods for the Early Diagnosis of Pancreatic Cancer

Abstract

This study reviews the research progress, current applications, and future challenges of artificial intelligence and statistical methods in the early diagnosis of pancreatic cancer. Dubbed the ‘king of cancers’ due to its insidious early symptoms and rapid metastasis, pancreatic cancer exhibits low early detection rates, with traditional CT, MRI, and biomarker diagnostics demonstrating limited sensitivity. Artificial intelligence demonstrates significant potential in deep learning, machine learning, multimodal data analysis, medical image recognition, and clinical decision support systems, substantially enhancing the detection rate of small lesions and diagnostic accuracy. By integrating multi-source medical imaging, genomic, and clinical data, AI not only improves screening efficiency but also assists clinicians in formulating personalised treatment strategies. However, current limitations include inadequate data standardisation, insufficient explainability, and insufficient external validation. Future efforts should rely on large-scale, multicentre clinical studies to advance the clinical translation and practical application of AI models in pancreatic cancer early screening, thereby providing new avenues for reducing mortality rates.