Deep Learning Improves the MAGGIC Risk Score in Predicting Contrast-Induced Nephropathy in ST Elevation Myocardial...

Area of Science:

• Cardiology
• Nephrology
• Artificial Intelligence in Medicine

Background:

• Contrast-induced nephropathy (CIN) is a significant complication following primary percutaneous coronary intervention (pPCI) in ST-elevation myocardial infarction (STEMI) patients.
• Early identification of patients at high risk for CIN is critical for improving outcomes and reducing mortality.
• The utility of the Meta-Analysis Global Group in Chronic Heart Failure (MAGGIC) risk score for predicting CIN requires further investigation.

Purpose of the Study:

• To evaluate the effectiveness of deep learning (DL) models incorporating the MAGGIC score and clinical parameters for predicting CIN in STEMI patients undergoing pPCI.
• To compare the performance of various DL models against traditional machine learning algorithms for CIN prediction.

Main Methods:

• A retrospective analysis of 1403 STEMI patients treated with pPCI.
• Development and comparison of DL models (multilayer perceptrons, TabNet, TabTransformer, KAN) and logistic regression using the MAGGIC score and 21 clinical parameters.
• Utilized SHapley Additive exPlanations (SHAP) for predictor identification.

Main Results:

• The Kolmogorov-Arnold Networks (KAN) model demonstrated superior performance with an Area Under the Curve (AUC) of 0.92 for CIN prediction.
• Key predictors identified by SHAP analysis included pain-to-balloon time, contrast volume, baseline creatinine, and the MAGGIC score.
• Patients who developed CIN exhibited higher mortality, prolonged hospital stays, and more comorbidities.

Conclusions:

• Combining MAGGIC risk scoring with DL models, especially KAN, significantly enhances CIN prediction in STEMI patients undergoing pPCI.
• This advanced predictive approach facilitates early identification of high-risk individuals, enabling timely implementation of nephroprotective strategies.
• The developed models and potential web-based calculator can aid clinical decision-making for CIN prevention.