Unveiling Key Biomarkers of Cardiovascular Risk in Psoriasis Through Explainable Artificial Intelligence.

Area of Science:

• Cardiology
• Dermatology
• Artificial Intelligence

Background:

• Psoriasis patients have a higher risk of cardiovascular diseases (CVD).
• Accurate CVD risk prediction is crucial for early intervention in this population.

Purpose of the Study:

• To develop and validate a machine learning model for predicting CVD risk in psoriasis patients.
• To identify key clinical and biochemical predictors of CVD in this cohort.

Main Methods:

• Utilized clinical and biochemical data from 2685 psoriasis patients.
• Employed a nested cross-validation framework with randomized hyperparameter search for model evaluation.
• Applied the Boruta algorithm for feature selection and Explainable AI (SHAP, LIME, Anchors) for model interpretability.

Main Results:

• The CatBoost model demonstrated superior performance (OOF ROC-AUC = 0.908).
• Key predictors included age, systolic blood pressure, apolipoprotein B, and fasting blood glucose.
• Explainable AI confirmed that older age, elevated blood pressure, and metabolic dysregulation are strong CVD risk factors.

Conclusions:

• Explainable AI can significantly improve CVD risk stratification in psoriasis patients.
• The developed model serves as a scalable tool for preventive cardiology.
• Future research should focus on external validation and diverse cohorts.