Enhanced classification of medical data using class level influence probability

Abstract

The modern society has higher influence of various diseases on human beings who are habituated to living in indifferent systems of life. Hence, they are exposed to different threats of dangerous diseases. The medical practitioners have the knowledge of most diseases and their related symptoms which are more similar to other diseases and there will be common symptoms that can be identified at different disease classes. The medical practitioners identify the class of disease, based on the symptoms and the number of instances. However, the previous records make the conclusion concrete one. Whatever be the case, the history makes the decision very supportive and the decisive support system needs huge number of data for concrete study. To support the decisive support systems, the medical data sets are used for several purposes. The medical data sets are higher dimensional one, as they combine the personal, medical, diagnosis, treatment and other related information. The researcher has identified different issues through the newlineanalysis of classification problem and the major one is the consideration of features. The previous algorithms use only limited number of features in measuring the similarity between the data points. This introduces higher false ratio because, the diseases would have similar symptoms. Hence, it is identified with more number of features which have to be considered. Besides, any disease would have specific influence from certain feature which encourages the disease to occur on the person or patient. newline newline