Early prediction of the evolution and prognosis of mild acute pancreatitis through Dual Source-CT and Formal Methods

Acute Pancreatitis (AP) is a common gastrointestinal disease with variable clinical outcomes, ranging from mild, self-limited cases to severe episodes that can be life-threatening. Early identification of patients at risk of progression from Mild Acute Pancreatitis (MAP) to severe/necrotizing pancreatitis is critical for early intervention, reducing morbidity, and optimizing resource allocation. This study proposes a radiomic classification model aimed at stratifying patients with MAP according to their progression. In contrast to the state-of-the-art, Model Checking is used to address the problem of the explosion of states in mathematical models. We retrospectively analyzed radiological data from a cohort of patients with MAP, obtaining an average metric of 80% which indicates strong classification ability. The use of Formal Methods allows us to have greater interpretability of the results. This model-building strategy can serve as a valuable tool for medical specialists, providing early insights for personalized management plans, and for computer scientists, increasing knowledge of the disease in the scientific community.