Multi-cascaded heart disease prediction using hybrid deep learning and optimization techniques.

Area of Science:

• Artificial Intelligence
• Cardiology
• Machine Learning

Background:

• Heart disease remains a leading cause of mortality worldwide.
• Accurate and early prediction is crucial for effective patient management.
• Existing prediction models often face challenges with data complexity and feature selection.

Purpose of the Study:

• To propose a novel deep learning model for enhanced heart disease prediction.
• To introduce an optimized feature selection method for improved model performance.
• To validate the model's efficacy using a benchmark dataset.

Main Methods:

• Data preprocessing using NaN fill and data normalization.
• Optimal weighted feature selection via Mutated Iteration-based Fire Hawk with Coyote Optimization (MI-FHCO).
• Heart disease prediction using a Multi-Cascaded Deep Learning Network (MDLNet).

Main Results:

• The proposed deep learning model achieved a high accuracy rate of 96.65% on dataset 4.
• The MI-FHCO algorithm effectively identified optimal features for prediction.
• The MDLNet demonstrated superior performance in heart disease classification.

Conclusions:

• The novel deep learning approach offers a highly accurate method for heart disease prediction.
• The MI-FHCO feature selection technique enhances model specificity and sensitivity.
• This model shows significant potential for clinical application in early heart disease detection.