Temperature is a key parameter in organ preservation that has been consistently linked to primary nonfunction (PNF). In this communication, and for the first time anywhere, continued and direct measurements of human liver intraparenchymal temperatures are reported in six clinical cases of orthotopic liver transplantations (OLT). These measurements cover the entire transplantation procedure and include the full transport phase. In contrast with long-held beliefs, these data demonstrate that liver allograft temperatures reach and stabilize at near 0 degrees C, instead of 4 degrees C, during transport using standard protocols. Furthermore, these low temperatures do not appear to contribute to graft failure when negative factors such as long preservation, the presence of hepatic steatosis, or advanced donor age are present. The clinical and experimental implications of these findings, together with other relevant elements derived from the direct and continuous monitoring of human liver allograft intraparenchymal temperatures, are discussed.

Acute pancreatitis (AP) presents a significant clinical challenge due to its wide range of severity, from mild cases to life-threatening complications such as severe acute pancreatitis (SAP), necrosis, and multi-organ failure. Traditional scoring systems, such as Ranson and BISAP, offer foundational tools for risk stratification but often lack early precision. This review aims to explore the transformative role of artificial intelligence (AI) and machine learning (ML) in AP management, focusing on their applications in diagnosis, severity prediction, complication management, and treatment optimization. A comprehensive analysis of recent studies was conducted, highlighting ML models such as XGBoost, neural networks, and multimodal approaches. These models integrate clinical, laboratory, and imaging data, including radiomics features, and are useful in diagnostic and prognostic accuracy in AP. Special attention was given to models addressing SAP, complications like acute kidney injury and acute respiratory distress syndrome, mortality, and recurrence. AI-based models achieved higher AUC values than traditional models in predicting acute pancreatitis outcomes. XGBoost reached an AUC of 0.93 for early SAP prediction, higher than BISAP (AUC 0.74) and APACHE II (AUC 0.81). PrismSAP, integrating multimodal data, achieved the highest AUC of 0.916. AI models also demonstrated superior accuracy in mortality prediction (AUC 0.975) and ARDS detection (AUC 0.891) AI and ML represent a transformative advance in AP management, facilitating personalized treatment, early risk stratification, and allowing resource utilization to be optimized. By addressing challenges such as model generalizability, ethical considerations, and clinical adoption, AI has the potential to significantly improve patient outcomes and redefine AP care standards globally.