Integrated deep learning for cardiovascular risk assessment and diagnosis: An evolutionary mating algorithm-enhanced...

Area of Science:

• Cardiovascular research
• Artificial intelligence in medicine
• Deep learning for healthcare

Background:

• Cardiovascular diseases (CVD) are a leading global cause of mortality.
• Accurate and efficient predictive models are crucial for early detection and intervention.
• Existing predictive models require enhancement for improved accuracy and robustness.

Purpose of the Study:

• To develop and validate a dual-output deep learning model for cardiovascular disease (CVD) prediction.
• To optimize the model using the Evolutionary Mating Algorithm (EMA) for enhanced performance.
• To compare the EMA-optimized model against other optimization algorithms.

Main Methods:

• A hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) deep learning architecture was employed.
• The Evolutionary Mating Algorithm (EMA) was utilized for hyperparameter optimization.
• Performance was benchmarked against models optimized with Particle Swarm Optimization (PSO) and Barnacle Mating Optimization (BMO).

Main Results:

• The EMA-optimized dual-output CNN-LSTM model demonstrated superior predictive accuracy.
• Risk prediction metrics included MAE of 0.018, MSE of 0.0006, RMSE of 0.024, and R² of 0.98.
• The diagnostic task achieved 70% accuracy and 80% precision, outperforming PSO and BMO.

Conclusions:

• The Evolutionary Mating Algorithm (EMA) is effective for optimizing dual-output deep learning models in healthcare.
• The proposed model shows significant potential for improving cardiovascular risk stratification and early diagnosis.
• This approach offers a robust tool for clinical decision-making in cardiovascular medicine.