Automated Chronic Obstructive Pulmonary Disease (COPD) detection and classification using Mayfly optimization with deep belief network model

Chronic Obstructive Pulmonary Disease (COPD) is a progressive and debilitating respiratory condition affecting millions worldwide. Respiratory disease affects quality of life and poses a substantial economic burden on patients and families. Diagnosis of COPD is unreliable as the test depends on the effort made by the tester and testee. Routine healthcare data collection from patients enables the identification of COPD subtypes so that physicians can define the disease severity and progression. Selecting optimal features from a large volume of healthcare data increases the computation burden and may lead to misclassification. In this research work, the Mayfly optimization algorithm is used for optimal feature selection from the COPD Patients Dataset, and the Deep Belief Network is then used for classification. The proposed Mayfly Optimized Deep Belief Network (MODBN) performance is experimentally verified using a benchmark dataset, and the performances are comparatively analyzed with traditional machine learning algorithms. A maximum classification accuracy of 96.89 % was attained by the proposed model in the classification of COPD compared to traditional machine learning algorithms.