An Effective Approach to Detect Liver Disorder using CNN Algorithm in Comparison with Random Forest Algorithm to Measure Accuracy

Abstract

Aim: The ultimate aim of this work is to show the better mean accuracy of CNN algorithm in comparison with random forest algorithm on detection of liver disorder. Materials & Methods: For identification of effective approaches to detect liver disorder, the conventional neural network algorithm is used comparatively with the random forest algorithm which is an existing algorithm. For each group sample size is taken as 20 and total sample size is taken as 40. Sample size calculation was done using clincalc. com by keeping g-power at 80%, confidence interval at 95 % and threshold at 0.05 %. Result: By this innovative liver disorder detection method, it is concluded that the random forest algorithm is much lower than CNN algorithm in predicting the liver disorder. Each sample shows different accuracy values, the overall mean accuracy values shows that the CNN algorithm is having greater accuracy of 95.86% than the random forest algorithm which is having 93.88% on analysis of liver disorder approach. For CNN and random forest algorithms the statistical significance of different values is p=0.001 i.e., p<0.05 (Independent sample T-test) provided by statistical results. Conclusion: On analysis of liver disorder it is clearly known that the CNN algorithm shows better mean accuracy value then random forest algorithm. The CNN algorithm shows the accuracy of 95.86% and random forest shows the accuracy of 93.88%.