A hybrid stacked autoencoder and support vector machines-based expert system for heart failure detection

Area of Science:

• Medical Diagnostics
• Machine Learning
• Artificial Intelligence in Healthcare

Background:

• Heart failure (HF) detection is crucial for timely medical intervention.
• Existing methods for HF detection require improvement in accuracy and efficiency.
• Advanced computational approaches can enhance diagnostic capabilities for complex diseases like HF.

Purpose of the Study:

• To introduce a novel hybrid three-stage expert system for enhanced heart failure detection.
• To improve the accuracy and efficiency of heart failure diagnosis through an integrated computational approach.
• To provide a robust decision support tool for healthcare professionals in identifying heart failure.

Main Methods:

• Feature extraction using stacked autoencoder (AE).
• Feature selection via an [Formula: see text]-penalized support vector machine (SVM).
• Classification using a non-linear SVM with an RBF kernel.

Main Results:

• Achieved high accuracy (97.78%), sensitivity (97.56%), and specificity (97.96%).
• Demonstrated superior performance compared to current state-of-the-art methods.
• Validated using a benchmark heart failure dataset with metrics including MCC and AUC.

Conclusions:

• The proposed hybrid expert system offers a robust and efficient solution for heart failure detection.
• The system's ability to achieve high performance with a reduced feature set underscores its efficiency.
• This approach provides valuable decision support for healthcare professionals in diagnosing heart failure.