A Novel Model based on Non Invasive Methods for Prediction of Liver Fibrosis

Abstract

Abstract Serial liver biopsies are typically the gold standard for diagnosis of liver fibrosis progression. However, it is associated with serious complications, inconvenient to patients, in some cases it causes dying, and it is expensive. The spread and the danger of the hepatitis C virus which infect the liver leading to deadly diseases like liver fibrosis. The challenge is to substitute the liver biopsy (i.e. the invasive procedures) with non-invasive method depending on the computer added tools i.e. a decision support system. The proposed technique is making a rule mining depends on a complete search but not exhaustive to guarantee finding optimal itemset rules, to build a robust and precise decision support system, it depends on new pruning rules to efficiently reduce search and dimensional space. It is not only search 100% of the dataset but also finds all minimal unique rules. This introduces to an optimal itemset rules mining tool. It is employed to resolve this medical diagnosing issue with average accuracy 99.48% for 5-folds cross validation. This accuracy paves the way to utilize classification models as a clinically non-invasive and reliable method to assess the degree of liver fibrosis as a noninvasive method for prediction of liver fibrosis. This technique is applied using other datasets like Alzheimer Disorder as a biomarker extraction tool. Alzheimer’s infection (AD) is the most widely recognized neurodegenerative issue related to dementia in the elderly. Although, initiating events are still unknown, it is clear that AD results from a combining of genetic and environmental risk factors. Diagnosis can be improved by the use of biological measures. However, it takes time (Deterioration of patient condition), the challenge is to save time. The proposed technique is employed to resolve this issue with average accuracy 97.15% for 10-folds cross-validation.