The frontiers of intelligent health services: cardiovascular disease prediction using novel machine learning methods...

Area of Science:

• Medical Informatics
• Computational Biology
• Biostatistics

Background:

• Cardiovascular disease (CVD) is a leading cause of mortality and a significant factor in aging populations worldwide.
• Accurate CVD risk assessment is crucial for timely and effective early intervention strategies.
• Traditional risk assessment models can be enhanced through advanced computational techniques.

Purpose of the Study:

• To investigate the efficacy of integrating optimization algorithms with discriminant analysis models for improved CVD risk prediction.
• To compare the performance of different optimized models, specifically focusing on the LDGO model.

Main Methods:

• Utilized Linear Discriminant Analysis (LDA) and Quadratic Discriminant Analysis (QDA) as baseline models.
• Integrated optimization algorithms, including the Golf Optimization Algorithm (GOA) and Leader Harris Hawk's Optimization (LHHO), with LDA and QDA.
• Empirically evaluated the performance of the integrated models, particularly the LDGO model (LDA + GOA).

Main Results:

• The LDGO model demonstrated high predictive accuracy, achieving 0.948 in the training phase and 0.946 in the test phase.
• Integration with optimization algorithms significantly improved the performance of discriminant analysis models for CVD risk assessment.
• The LDGO model emerged as the most effective among the investigated models.

Conclusions:

• Optimized discriminant analysis models, particularly LDGO, offer a promising approach for enhancing cardiovascular disease risk prediction.
• These advanced models can facilitate more accurate early detection and intervention, potentially reducing CVD-related mortality and improving health outcomes in aging populations.