Developing an Interpretable Machine Learning Model for Early Prediction of Cardiovascular Involvement in Systemic...

Area of Science:

• Cardiology
• Rheumatology
• Artificial Intelligence

Background:

• Cardiovascular disease is a major cause of mortality in patients with systemic lupus erythematosus (SLE).
• Early detection of cardiac complications in SLE is crucial for improving patient prognosis and outcomes.
• This research focuses on identifying predictors and developing a predictive model for cardiac involvement in SLE.

Purpose of the Study:

• To identify key clinical and biological factors associated with cardiac involvement in SLE patients.
• To develop and validate an interpretable machine learning (ML) model for predicting the risk of cardiac complications in SLE.
• To enhance early identification and management of high-risk SLE individuals.

Main Methods:

• A retrospective analysis of 1,023 SLE patients hospitalized between 2000 and 2021.
• Utilized three feature selection techniques (Random Forest, LASSO, XGBoost) to identify predictive variables.
• Trained and tested seven ML algorithms, with Gradient Boosting Machine (GBM) selected for its performance and interpretability.

Main Results:

• Over a median follow-up of 3,737 days, 18.77% of patients developed cardiac involvement.
• Seven key predictors were identified: arthritis, hypertension, HDL-C, LDL-C, total cholesterol, CRP, and ESR.
• The GBM model achieved an AUC of 0.748 and an accuracy of 0.779.

Conclusions:

• This study presents the first interpretable ML model for predicting cardiac involvement risk in SLE.
• The developed GBM model demonstrates optimal performance and provides insights into decision-making processes for clinicians.
• The model facilitates early identification of SLE patients at high risk for cardiac complications.