PSO optimized 1-D CNN-SVM architecture for real-time detection and classification applications

Area of Science:

• Biomedical Engineering
• Artificial Intelligence
• Machine Learning

Background:

• Chronic Kidney Disease (CKD) diagnosis often relies on complex laboratory tests.
• There is a need for rapid, non-invasive, and accurate methods for early CKD detection.
• Saliva analysis presents a promising avenue for developing point-of-care diagnostic tools.

Purpose of the Study:

• To propose and validate a novel Particle Swarm Optimized (PSO) One-Dimensional Convolutional Neural Network with Support Vector Machine (1-D CNN-SVM) architecture.
• To enable real-time detection and classification of diseases, specifically Chronic Kidney Disease (CKD).
• To develop a hardware model for CKD detection using saliva samples.

Main Methods:

• A novel hardware model was designed to monitor urea concentration in saliva by converting it to ammonia using urease enzyme.
• Ammonia levels were measured using a semiconductor gas sensor, generating raw signal data.
• The sensor data was processed by a PSO-optimized 1-D CNN-SVM architecture for feature extraction and classification.

Main Results:

• The proposed PSO-optimized 1-D CNN-SVM architecture demonstrated superior performance compared to conventional methods.
• Optimal features were extracted directly from the raw sensor signal, reducing computational time and complexity.
• The system achieved a high accuracy of 98.25% in detecting Chronic Kidney Disease (CKD).

Conclusions:

• The integrated approach of sensor-based detection and PSO-optimized 1-D CNN-SVM offers an effective solution for real-time disease diagnosis.
• This method provides a computationally efficient and accurate approach for CKD detection from saliva.
• The developed architecture holds potential for non-invasive, point-of-care diagnostic applications.