Keen Ear

Abstract

In this modern world of today, everything is being automized. Data Science, Machine Learning and Artificial Intelligence are fields that are being exploited to make strides in almost all walks of life. A lot of jobs that were considered to be impossible without human effort are now being done not just automated programs but with minimum amount of processing too. These advantages of Data Science to extract the most useful information from the rawest of data, can and is being used to further the field of healthcare as well. It is no secret that respiratory diseases are some of the most life-threatening diseases of all. Five of the most common respiratory diseases are actually most common cause of overall deaths around the world. It is evident that an important part to the treatment of these respiratory diseases like asthma, bronchitis, COPD, URTI etc. is timely diagnosis. The faster the disease is diagnosed the faster it can be treated. The problem that is generally faced in this process is that different respiratory diseases have different diagnosis methods and the time taken in carrying these out can be very vital if used in the treatment of the patient. This project was development of a Machine Learning and Data Science project that will be able to reduce this extra time that is used up in diagnosis of the patient’s exact disease and which if used for the patient’s treatment can be extremely vital. This thesis looks at the process of development of an ML model that is able to tell a patient’s respiratory disease after listening to the respiratory audio of the patient. This project also produces a tangible device that employs said ML model to take the respiratory audio of a patient in live time and process the audio to extract important data from it and using that data the ML model would be able to classify the patient’s respiratory diseases. Thus, giving a one-stop respiratory disease diagnosis option to doctors and patients alike. The programming practices and concepts used in this project are Machine Learning, Data Science, Data Analysis, Microprocessor and Embedded Programming.