A Mobile Application for Early Diagnosis of Pneumonia in the Rural context

Abstract

Pneumonia is a respiratory infection resulting in inflammation of the lungs. The causes of this infectious disease could be attributed to viruses, bacteria or fungi. One of the many ways of detecting the disease is by a chest X-ray of the patient. The rural population in developing nations have limited access to doctors, medical diagnostic facilities, and hospitals. Hence, diagnosis is delayed resulting in adverse consequences. This paper is an attempt to design and develop a smartphone-based application (app) for the preliminary detection of pneumonia using X-ray images. The app is based on machine learning which identifies pneumonia, using a chest X-ray image of a patient with a 'MobileNets' model, trained on thousands of X-ray images of known cases of pneumonia. The app has been developed on Android Studio, incorporating TensorFlow library. The patient's chest X-ray is scanned and uploaded to the app using the smartphone camera. Additionally, an e-diagnosis facility is integrated into the app where qualified medical practitioners' advice is taken on the obtained results. A breathing pattern recorder module is developed, which, in future, could be integrated into the smartphone app to increase its accuracy in prediction.