Prediction of cardiovascular diseases based on GBDT+LR

Area of Science:

• Computational biology
• Medical informatics
• Machine learning in healthcare

Background:

• Cardiovascular diseases (CVDs) affect over 300 million people in China, with aging populations exacerbating the burden.
• Accurate and efficient CVD risk prediction is essential for disease prevention and public health management.

Purpose of the Study:

• To develop and evaluate a novel hybrid machine learning model for predicting cardiovascular disease risk.
• To enhance the predictive capabilities for cardiovascular disease by combining Gradient-Boosting Decision Trees (GBDT) and Logistic Regression (LR).

Main Methods:

• A hybrid model integrating GBDT and LR was developed, using GBDT's predictions as input features for the LR model to handle non-linear data.
• The proposed GBDT+LR model was evaluated against Logistic Regression (LR), Random Forest (RF), and Support Vector Machine (SVM) using the UCI cardiovascular disease dataset.
• A cardiovascular disease analysis and prediction platform was built using Spark, Vue, and SpringBoot frameworks.

Main Results:

• The GBDT+LR model demonstrated superior performance across multiple evaluation metrics, including accuracy, precision, specificity, F1-score, Matthews Correlation Coefficient (MCC), Area Under the Curve (AUC), and Area Under the Precision-Recall Curve (AUPR).
• Experimental comparisons confirmed that the GBDT+LR approach significantly outperformed traditional LR, RF, and SVM models in predicting cardiovascular disease risk.
• The developed platform successfully implemented the GBDT+LR algorithm for real-time cardiovascular disease risk probability prediction.

Conclusions:

• The hybrid GBDT+LR model offers the best prediction performance for cardiovascular disease risk assessment.
• This approach effectively addresses the limitations of LR in handling complex, non-linear relationships within medical data.
• The integrated platform provides a robust solution for analyzing and predicting cardiovascular disease risk, contributing to improved public health strategies.