Abstract
COVID-19 diagnosis is integral to the efforts of limiting the spread of the disease. However, diagnosing COVID-19 poses challenges and risks of further spread of the disease whilst also not always being easily accessible to all members of the public. This paper proposes a method to develop a novel non-invasive method to automatically classify the most common symptoms of COVID-19. The method is based on the classification of sounds associated with coughing and shortness of breath of potential COVID-19 patients. Our solution is based on Deep Learning technology through Convolutional Neural Network (CNN) with the aim of recognizing the shortness of breath and dry cough of carriers of COVID-19. This system will provide an effective and objective method for diagnosis to enable healthcare providers and the public to assess their conditions with high confidence prior to proceeding with further investigation and treatments. The core of this e-health based research focuses on developing an advanced AI based classification diagnostic engine with high accuracy that feeds into different modules of the platform according to the need-analysis which will be conducted. The classifier will also be trained incrementally through innovative feedback channels that rely on the consensus of the expertise labelling to ensure system integrity. The outcome of this research will be a platform accessible by patients, healthcare workers and decision makers through mobile computing technologies and remote terminals that will be used to capture the COVID-19 audible symptoms as well as presenting the outcome.