The clinical course of acute pancreatitis (AP) can be variable depending on the severity of the disease, and it is crucial to predict the probability of organ failure to initiate early adequate treatment and management. Therefore, possible high-risk patients should be admitted to a high-dependence unit. For risk assessment, we have three options: (1) There are univariate biochemical markers for predicting severe AP. One of their main characteristics is that the absence or excess of these factors affects the outcome of AP in a dose-dependent manner. Unfortunately, all of these parameters have low accuracy; therefore, they cannot be used in clinical settings. (2) Score systems have been developed to prognosticate severity by using 4-25 factors. They usually require multiple parameters that are not measured on a daily basis, and they often require more than 24 h for completion, resulting in the loss of valuable time. However, these scores can foresee specific organ failure or severity, but they only use dichotomous parameters, resulting in information loss. Therefore, their use in clinical settings is limited. (3) Artificial intelligence can detect the complex nonlinear relationships between multiple biochemical parameters and disease outcomes. We have recently developed the very first easy-to-use tool, EASY-APP, which uses multiple continuous variables that are available at the time of admission. The web-based application does not require all of the parameters for prediction, allowing early and easy use on admission. In the future, prognostic scores should be developed with the help of artificial intelligence to avoid information loss and to provide a more individualized risk assessment.
Keywords: BISAP; CRP; WBC; albumin; machine learning; mortality; severity.