Minireviews
Copyright ©The Author(s) 2021.
World J Hepatol. Dec 27, 2021; 13(12): 1977-1990
Published online Dec 27, 2021. doi: 10.4254/wjh.v13.i12.1977
Table 1 Review of articles where artificial intelligence has been studied in the context of non-alcoholic liver disease
Ref.
Dataset
Number
ML algorithms
Problem
Performance measures
Byra et al[15], 2018Department of Internal Medicine, Hypertension and Vascular Diseases, Medical University of Warsaw, Poland55Deep CNNAutomatically diagnose the amount of fat in the liver from US imagesAUROC, Delong statistical test, lasso regression method, Spearman correlation coefficient, Meng test
Perveen et al[16], 2018CPCSSN667907Decision treeClassification, NAFLD progression riskMicro- and Macro-average of Precision, Recall and F-measure, MCC, AUROC
Ma et al[17], 2018First Affiliated Hospital, College of Medicine, Zhejiang University, China10508Several, Weka open source softwareClassification, feature selectionAccuracy, specificity, precision, recall (i.e. sensitivity), and the F-measure
Vanderbeck et al[18], 2014Medical College of Wisconsin, Milwaukee, United States59SVMAutomated assessment of histological features of NAFLDPrecision rate, recall rate, and AUROC
Meffert et al[68], 2014SHIP4222Boosting algorithm, discrimination and calibration plotsScoring system for hepatic steatosis riskDiscrimination (AUROC) and calibration
Sowa et al[69], 2014University Hospital Essen82Logistic regression, decision trees, SVM, RFDistinguish NAFLD from ALDSensitivity, specificity, and accuracy
Kuppili et al[70], 2017Instituto Superior Tecnico, University of Lisbon, Portugal63Extreme Learning Machine- SLFFNNStratification of FLD disease in US liver imagesAUROC, reliability and stability analysis
Sorino et al[71], 2020MICOL cohort2970SVMStratify NAFLD risk to reduce need for imagingAccuracy, variance, calculated confidence limits (95%), the weight of each model (as a %) and the number of ultrasound examinations it could avoid
Wu et al[72], 2019New Taipei City Municipal Hospital Banqiao Branch577ANN, NB, RF, LRDiagnosis and risk stratification in NAFLDAccuracy, sensitivity, specificity
Table 2 Review of recently published studies where artificial intelligence-based algorithms have been applied to liver transplantation
Ref.
Dataset
Number
ML algorithms
Problem
Performance measures
Bertsimas et al[62], 2019STAR dataset-OCTPredict 3 mo waitlist mortality-OPOMROC curve
Cruz-Ramírez et al[63], 2013Spanish multi-center study-Radial basis function NNImprove donor-recipient matching using rule-based allocation—MPENSGA 2 algorithmAccuracy, minimum sensitivity, ROC curve, RMSE, Cohen’s kappa
Briceño et al[64], 2014Spanish multi-center study1003Neural Net Evolutionary ProgrammingImprove equity in donor-recipient matchingMultiple regression analysis, simple logistic regression analysis, ROC curve
Ayllón et al[73], 2018King’s College Hospital,United Kingdom + MADR-E, Spain1437ANNClassification, end-point (3 mo, 1 yr)ROC curve
Wadhwani et al[74], 2019UNOS1482RFClassification, end-point (3 yr)Chi-square test, t-test, Wilcoxon rank sum test
Dorado-Moreno et al[75], 2017King’s College Hospital, United Kingdom + MADR-E, Spain1492Ordinal ANNOrdinal classification, fourclassesMAE and the MZE, accuracy, GMS, AMAE
Guijo-Rubio et al[76], 2019UNOS39095Cox, SVM, GBSurvival timeC-index, ROC curve, concordance index ipcw
Lee et al[77], 2018Seoul National University Hospital1211Several ML methods compared, GBM found to be bestPrediction of AKI after liver transplantROC curve, accuracy
Lau et al[78], 2017Austin Hospital, Melbourne, Australia180RF, ANN, logistic regressionPredict 30-d risk of graft failureROC curve