Prediction of coronary heart disease in gout patients using machine learning models

Area of Science:

• Cardiology
• Medical Informatics
• Machine Learning

Background:

• Gout patients face a higher risk of cardiovascular diseases, particularly coronary heart disease (CHD).
• Clinical screening for CHD in gout patients remains a significant challenge, leading to potential missed diagnoses or excessive testing.

Purpose of the Study:

• To develop a machine learning-based diagnostic model for predicting CHD in gout patients.
• To enhance the accuracy and efficiency of CHD screening in this high-risk population.

Main Methods:

• A binary classification model was created using data from over 300 patients, divided into gout and gout+CHD groups.
• Eight clinical indicators were selected as features for machine learning algorithms.
• Eight models, including logistic regression, SVM, random forest, and XGBoost, were evaluated, employing a combined sampling technique for imbalanced data.

Main Results:

• Stepwise logistic regression and SVM demonstrated superior AUC values.
• Random forest and XGBoost models exhibited excellent recall and accuracy.
• Key high-risk factors were identified as effective predictors for CHD in gout patients.

Conclusions:

• Machine learning models show promise in accurately predicting CHD risk in gout patients.
• The identified high-risk factors can inform clinical practice for better CHD screening strategies.
• This approach aims to optimize diagnostic pathways, reducing both missed diagnoses and unnecessary examinations.