Design of application-oriented disease diagnosis model using a meta-heuristic algorithm.

Area of Science:

• Healthcare technology
• Internet of Things (IoT)
• Artificial Intelligence in Medicine

Background:

• Healthcare relies on technology for diagnosis and treatment, with IoT offering remote patient monitoring solutions.
• IoT generates vast patient data, necessitating efficient analysis for timely diagnosis and care.
• Existing diagnostic methods struggle with big data, imbalanced datasets, and overfitting.

Purpose of the Study:

• To introduce a meta-heuristic optimization method for analyzing extensive IoT data for patient health monitoring.
• To enhance patient safety monitoring through advanced data analysis techniques.

Main Methods:

• Utilized particle swarm optimization (PSO) to optimize data for a diabetes diagnosis model.
• Employed convolutional neural networks (CNN) for illness prediction.
• Integrated PSO with CNN (PSO-CNN) for comprehensive data analysis.
• Used Support Vector Machine for cardiac risk prediction based on diabetes data.

Main Results:

• The PSO-CNN model achieved 92.6% accuracy, 92.5% precision, and 93.2% recall for diabetic disease prediction.
• The model demonstrated a 94.2% F1-score and a 4.1% quantization error.
• The approach shows significant improvements in predictive performance for healthcare diagnostics.

Conclusions:

• The developed PSO-CNN approach effectively predicts diabetic disease and cardiac risk.
• This method offers a robust solution for big data analysis in remote patient monitoring.
• The approach has potential applications in identifying other conditions, such as cancer cells.