Intelligent cardiovascular disease diagnosis using deep learning enhanced neural network with ant colony...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Computational Biology

Background:

• Cardiovascular disease (CVD) prevalence is increasing globally, necessitating early and precise diagnosis.
• Diagnosing CVD is challenging due to complex healthcare datasets and the difficulty in identifying feature connections for accurate predictions.
• Deep Learning (DL) and Machine Learning (ML) show promise for analyzing large medical datasets but face challenges in training and validation accuracy.

Purpose of the Study:

• To develop an advanced model for intelligent cardiovascular disease diagnosis (ICVD-ACOEDL).
• To enhance the accuracy and efficiency of CVD classification using feature selection and hyperparameter optimization.
• To address the limitations of existing methods in handling complex medical data for reliable CVD detection.

Main Methods:

• Data pre-processing using min-max scaling for consistency.
• Feature selection (FS) via Ant Colony Optimisation (ACO) to identify optimal feature subsets.
• Hyperparameter optimization of a deep learning enhanced neural network (DLENN) classifier using Bayesian optimization.

Main Results:

• The ICVD-ACOEDL model demonstrated superior performance compared to existing techniques on benchmark medical datasets.
• ACO-driven feature selection significantly enhanced the performance of the DLENN classifier.
• Bayesian optimization effectively tuned DLENN hyperparameters, improving CVD classification accuracy.

Conclusions:

• The ICVD-ACOEDL model offers a robust solution for improving CVD classification efficiency and accuracy.
• The integration of ACO for FS, min-max scaling, and Bayesian optimization provides a workable approach for real-world medical applications.
• This intelligent diagnostic system has the potential for significant impact on cardiovascular disease diagnosis and patient outcomes.