Disease diagnosis using machine learning: A comparative study

Abstract With time new diseases are coming up, there are diseases where the symptoms are considered to be harmless in 90% of the cases like cold, sneezing, and fever. People usually avoid going to the doctor for such symptoms and prefer taking some normal medicines like paracetamol. Such medicines won’t cure the diseases, and if not treated in time, it can be dangerous for the patient. To help provide people with the diagnoses of the disease, this chapter presents a comparative study of disease diagnosis that would require the sufferers to input the basic symptoms they are facing and would inform them about the disease they may be suffering from along with the probability of other diseases. The evaluation of the algorithm developed will help in the diagnosis of the diseases so that proper care can be taken. The developed model contains symptoms that would be displayed on the screen in groups of 10 starting from mere mild symptoms to serious ones, the user will have to select the appropriate symptoms, and the diagnosis prototype analyzes various combination of the symptoms occurring together through the machine learning model and displays the most probable disease diagnosed. The prototype proposed uses decision tree classification and deep neural networks to classify symptoms of diseases. The classification makes it easier for the user to identify the disease he/she is suffering from. It further compares the results of both techniques.