A cough type chronic disease prediction scheme using machine learning and diagnosis support system using a mobile application

Machine learning has been found to considerably lower the probability of inaccurate diagnoses when incorporated into modern diagnostic procedures. Different from the literature works, this paper proposes a method for diagnosing the three most similarly symptomised cough-type chronic diseases: COPD, bronchial asthma, and pneumonia. The symptoms of cough-type chronic disease patients admitted to the hospital were collected from eight medical colleges spread out over Bangladesh to construct the classifying model. This paper proposes a set of 27 attributes for appropriately classifying cough-type chronic disease instances. Several machine learning methods are tested using the dataset. Our research suggests gradient tree boosting to be the most effective, with a classification accuracy of 91%, despite the fact that previous research has identified support vector machine and random forest to be the most efficient models in these kinds of classification tasks. This paper also developed a mobile application for the diagnostic support systems.