Table 1.

Comparative analysis of existing work on hepatitis C virus prediction.

References	Used techniques	Performance metrics used	Data set and number of instances	Outcomes	Challenges
45	RF, SVM, GB	Precision, accuracy, miss rate	Online UCI Dataset, 668 instances	RF achieves 89% precision and a 17.2% miss rate	Cannot predict beyond the range of training data and overfitting issues
12	RF, KNN	Precision, accuracy, recall, F-measure, confusion matrix	A laboratory examination dataset with 200 instances	RF achieves 6% better results than KNN and other methods	It performs better for limited data
13	SVM, DT, GB, LR, NB, KNN, XGB, RF	Accuracy	Online Kaggle Dataset, with 4462 instances	The DB methods show 91.2 & accuracy over other methods	Fewer data samples utilize
14	RF	Precision	UCI data set with 670 instances	RF achieves 89.6% precision over other ML methods	Data inconsistency issues
15	RF, SVM	Precision, the miss rate	NCA Hospital dataset with 425 instances	RF achieves 87.6% precision	It performs better for limited data
16	SVM, ANN, KNN	Precision, accuracy, recall	Online UCI dataset with 295 instances	ANN Achieves 90.1% precision in training and testing	Limited parameters were considered in the experimental analysis
17	SVM, RF, DT, BN, NN, NB	Precision, recall, F-measure, detection rate, and recall	Online Kaggle dataset with 559 Instances	NN performs better and achieves more than 11.6% better results than other methods	It performs better for limited data
18	Extreme learning machine	Precision, miss rate	Online Kaggle dataset with 550 instances	Better precision as compared to the SVM method	Limited parameters
19	ANN	Accuracy, miss rate	Lahore Hospital dataset with 289 instances	ANN achieves better results in terms of accuracy and miss rate %	Data inconsistency issues
20	PSO, GA, REP, DT- C4.5 and CART, ADT, MLR, RT	Precision, recall, accuracy, miss rate	Egypt HCV dataset, with 669 instances	GA methods show better classification outcomes	It performs better for limited data
21	SVM, NB, NN, DT	Accuracy and miss rate %	Online UCI dataset with 335 instances	NN achieves better accuracy and misses rate%	Limited parameters were considered in the experimental analysis
22	SVM, simulated annealing (SA)	Sensitivity, specificity, precision, and accuracy	Online Kaggle dataset with 295 instances	SVM achieves better results than existing ML methods	Data inconsistency issues
23	Binary LR	TPR and accuracy	Online UCI dataset with 269 instances	Binary LR achieves better TPR and accuracy %	Performs better for limited data
24	ANN, NN, and SVM	Precision and accuracy	Online UCI dataset with 295 instances	ANN achieves better precision	Limited parameters were used