An Artificial Neural Networks Model for Early Predicting In-Hospital Mortality in Acute Pancreatitis in MIMIC-III

Area of Science:

• Medical informatics
• Artificial intelligence in healthcare
• Clinical prediction models

Background:

• Early and accurate assessment of acute pancreatitis (AP) severity and prognosis is crucial for patient outcomes.
• Developing predictive models at admission can significantly improve patient management.
• This study focused on creating an artificial neural network (ANN) for early in-hospital mortality prediction in AP.

Purpose of the Study:

• To develop and evaluate an artificial neural network (ANN) model for predicting in-hospital mortality in patients with acute pancreatitis (AP).
• To identify key risk factors associated with mortality in AP patients.
• To compare the performance of the ANN model against traditional scoring systems.

Main Methods:

• Utilized data from the Medical Information Mart for Intensive Care-III (MIMIC-III) database for patients diagnosed with AP.
• Employed a backpropagation artificial neural network (ANN) approach using clinical and laboratory data.
• Performed multivariate logistic regression analysis to identify independent risk factors for mortality.

Main Results:

• Analyzed 337 AP patients, with an in-hospital mortality rate of 11.2%.
• The ANN model achieved an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.769, outperforming logistic regression (0.607), Ranson score (0.652), and SOFA score (0.401).
• Identified three independent risk factors for in-hospital mortality through logistic regression.

Conclusions:

• This study presents the first ANN predictive model for in-hospital mortality in AP patients using the MIMIC-III database.
• The developed ANN model demonstrates superior predictive performance compared to existing methods.
• The model facilitates early screening of AP patients at high risk of fatal outcomes, enabling prompt clinical intervention.