ARCHIVES

Abstract: Nowadays, every kind of data is digital data. Healthcare is so advanced that physiological data is stored as digital data once it is created. However, diagnosing diseases is still done by clinicians manually. But Machine Learning can be helpful in this situation. ML has a wide variety of applications which will be useful to detect any patterns of certain diseases within patient electronic healthcare records and it is not subjected to any past experiences like human’s. While solving a problem with Machine Learning, we have to create different models and select the best among them. There are plenty of machine learning or deep learning classes we can try.As there’s many algorithms and neural network architectures, we have to select the most promising and advanced model first. Before selecting the model , we have to select our data. Medical data has a lot of privacy concerns, so it is not widely available. Also the format and quality of data do arise other challenges like effort to clean and preprocessing. Our Aim is to select the best data available and build models for diagnosing . First , we are focusing on brain tumors and then on Arrhythmia as these abnormalities should be detected earlier and as human inspection on these data, specifically MRI images depends strongly on their experience. Also, operator-assisted classification methods are not practical for large amounts of data and also are not reproducible. Therefore, it is highly desirable to use Machine Learning or computer- aided technologies to address these problem.