Photo Credit: wildpixel

The following is a summary of “Interpretable machine learning model for new-onset atrial fibrillation prediction in critically ill patients: a multi-center study,” published in the October 2024 issue of Critical Care by Guan et al. 

New-onset atrial fibrillation, the most common arrhythmia in patients who are critically ill, is linked to poor prognosis and high disease burden, highlighting the need for early risk identification. 

Researchers conducted a retrospective study to develop and validate a machine learning model to predict new-onset atrial fibrillation in patients who are critically ill. 

They used MIMIC-IV for training and a MIMIC-III subset for validation, applied LASSO regression for feature selection, and built the model with 8 machine learning algorithms. SHapley Additive exPlanations (SHAP) visualized model features and predictions. 

The results showed that among 16,528 patients with MIMIC-IV, 1,520 (9.2%) developed atrial fibrillation post-ICU admission. A model with 23 variables was built, with XGBoost performing best, achieving an AUC of 0.891 (0.873–0.888) in validation and 0.769 (0.756–0.782) in external validation. Key predictors included age, mechanical ventilation, urine output, sepsis, blood urea nitrogen, percutaneous arterial oxygen saturation, continuous renal replacement therapy, and weight. A risk probability greater than 0.6 was defined as high risk. A user-friendly interface was developed for clinician use. 

The study concluded that the developed machine learning model effectively predicts the risk of new-onset atrial fibrillation in patients who are critically ill, enhancing interpretability through SHAP, which aids clinicians in understanding causes and improving patient outcomes. 

Source: ccforum.biomedcentral.com/articles/10.1186/s13054-024-05138-0