| Popis: |
Background:Inflammation of liver causes acute or chronic infection called hepatitis. Inflammation is a sort of swelling that happens when body tissues are contaminated or injured. Liver is a vibrant organ of our body that is used to process nutrients, scrimmage infections and purify the blood. Heavy use of ethanol, some medication and toxin can cause hepatitis. There are multiple viral classifications of hepatitis that are hepatitis A, B, C, D and E. Hepatitis A, B and C are most common with different virus is responsible for each type of hepatitis. Methods: Data mining is the discovery of huge datasets to explore hidden patterns that is tough to detect with typical techniques. In medical field, data mining is very prominent with the importance of providing diagnosis and a very deep understanding of medical data. In the past, for certain disease, medical and family history, risk factors and physical examination were assessed by the test results. It can be said that a test can detect the disease but now machine learning offers multiple classifiers that are very helpful for the detection of diseases. The medical industries collect a lot of data that is not mined properly and as a consequence there is not an ideal use. To find out such hidden patterns and relationship between them often goes spoiled. The problem is how to analyze the data. Various techniques are used for data mining such as classification, neural networks for data finding and arrangement, clustering and association rule mining. The algorithm used in this study comes in supervised learning whose labels are specified, in which classification is the most esteemed technique that is used for several beneficial applications like artificial intelligence and beyond. A lot of Weka classifiers are applicable such as rules, trees, misc, meta, bayes and functions with their strengths and weaknesses. Thisstudy focus on Intelligent Clinical Decision Support System by using decision trees. Decision trees provide easier and simpler hierarchical structure which will help doctors in the medical field in which data collected by health care evaluation samples through machine learning in hepatitis research and development. At present, it is used for quick data analysis from learning data set obtained from the medical record of affected people. Now, learning capacity is well suited for clinical data analysis, and completes medical decisions, especially a big job in medical industry. So, to help and recognize effected patients to get a better evaluation of the resulting or derived classifier speed, accuracy and consistency of treatment can be used. Also, by inspiring the world-wide rapidly increase mortality of hepatitis, patients and the providing ability of large amount of victims, the aim of this study is to aid medical professionals in the diagnosis of hepatitis disease by using decision tree classifiers. Particularly, we have used Classification and Regression Tree (CART) and Java 48 (J48) algorithms by using 10-fold cross validation method to monitor and detect hepatitis disease with comprehensive medical accuracy. This research tries to enhance the learning process actively as well as to help the doctors. Results: From the comparison and investigation of the results revealed that J48 decision tree algorithm shows improved performance over CART algorithm. J48 predicted efficient results with highest classification rate and gives better understanding regarding performance parameters as compared to CART with an accuracy of 80% and sensitivity of 88% and specificity of 52% that will help physicians. Conclusion: This paper concluded that it may also be used to build an automated system that will help doctors for accurate diagnosis of chronic or severity of hepatitis disease detection. This decision support system has a very great potential to be further improved in the future. |
