Abstract
Hepatic cirrhosis represents an advanced stage of fibrosis in the liver, resulting from various conditions such as hepatitis and chronic alcoholism. This article uses data from the Mayo Clinic’s clinical trial on primary bile cirrhosis carried out between 1974 and 1984. Rapid and accurate identification of the condition is crucial for the implementation of effective interventions, capable of mitigating liver damage and preventing complications, especially in the early stages of the disease. The focus of the research is the evaluation of Data Mining (DM) techniques, using clinical data, to predict survival outcomes in patients with cirrhosis. The study uses the CRISP-DM methodology, analysing different classification algorithms, including k-Nearest Neighbours, Random Forest, Decision Trees, Gradient Boosting and Naive Bayes. The effectiveness of the model is evaluated, highlighting the Random Forest with Holdout Sampling and a ratio of 0.7, whose performance averages around 80%. The nominal attributes were also analysed in order to discover possible patterns of association.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have