PSO optimized 1-D CNN-SVM architecture for real-time detection and classification applications. (May 2019)

• Record Type: Journal Article
• Authors: Navaneeth, Bhaskar; Suchetha, M.
• Is Part Of: Computers in biology and medicine. Volume 108(2019)
• Page Start: 85
• Page End: 92
• Publication Date: 2019-05
• Subjects: Convolutional neural network -- Support vector machine -- Chronic kidney disease -- Salivary diagnosis -- Particle swarm optimization; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2019.03.017
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: In this paper, we propose a novel Particle Swarm Optimized (PSO) One-Dimensional Convolutional Neural Network with Support Vector Machine (1-D CNN-SVM) architecture for real-time detection and classification of diseases. The performance of the proposed architecture is validated with a novel hardware model for detecting Chronic Kidney Disease (CKD) from saliva samples. For detecting CKD, the urea concentration in the saliva sample is monitored by converting it into ammonia. The urea on hydrolysis in the presence of urease enzyme produces ammonia. This ammonia is then measured using a semiconductor gas sensor. The sensor response is given to the proposed architecture for feature extraction and classification. The performance of the architecture is optimized by regulating the parameter values using a PSO algorithm. The proposed architecture outperforms current conventional methods, as this approach is a combination of strong feature extraction and classification techniques. Optimal features are extracted directly from the raw signal, aiming to reduce the computational time and complexity. The proposed architecture has achieved an accuracy of 98.25%. Highlights: Developed an innovative 1-D CNN-SVM architecture for real-time detection and classification of CKD non-invasively. A novel sensing model capable of detecting CKD from the saliva sample has been developed. The Particle Swarm Optimization algorithm is introduced to enhance the classifier performance. Evaluated the ROC plot and computational time of the architecture for ascertaining the diagnostic ability of the model. The proposed model outperforms conventional data classification methods.