Heart disease detection based on internet of things data using linear quadratic discriminant analysis and a deep...

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Heart disease poses a significant global health challenge, with current diagnostic tools often lacking efficacy.
• There is a critical need for advanced, intelligent systems to accurately diagnose heart conditions.

Purpose of the Study:

• To design and develop an intelligent heart disease diagnosis application leveraging Internet of Things (IoT) sensor data and deep learning.
• To improve the accuracy and performance of heart disease diagnosis compared to existing methods.

Main Methods:

• Utilized IoT sensor data from the UCI machine learning repository and Cleveland Clinic Foundation for training and testing.
• Applied K-means for data denoising and clustering, followed by Linear Quadratic Discriminant Analysis for feature extraction.
• Developed and implemented a deep graph convolutional network (DG_ConvoNet) for heart disease classification and prediction.

Main Results:

• The DG_ConvoNet model achieved a diagnostic accuracy of 96%.
• Key performance metrics include 80% sensitivity, 73% specificity, 90% precision, and a 79% F-Score.
• The model demonstrated a 75% area under the Receiver Operating Characteristic (ROC) curve.

Conclusions:

• The proposed deep learning-based application effectively diagnoses heart diseases using IoT sensor data.
• The system demonstrates superior performance, offering a promising advancement in cardiac diagnostics.