Table 1

Current status of published articles on machine learning in axSpA

GoalsYear of publicationAuthorJournalInput dataMachine learning technologyPredictive effectiveness
Precision medicine
Early screening and diagnosis20207Deodhar et alClin RheumatolMedical claims database medical recordsOptimise the linear regression model A/BModel A/B PPV: 6.24% Simplified linear regression model PPV: 2.55%
20229Zhu et alRheumatol TherLaboratory indicatorsLASSO, SVM-RFE, RF and NomogramsTraining cohort AUC: 0.87 Validation cohort AUC: 0.82
202210Ye et alRheumatologyMRI and medical recordsmRMR, LASSO, multivariable logistic regression analysis and NomogramsRad-score:training/validation cohort AUC: 0.82. The clinical-radiomics nomogram model: training/validation cohort AUC: 0.9
202211Wen et alFront GenetGEO databaseLASSO, SVM-RFE, RF and NomogramsAUC>0.84
202212Han et alFront ImmunolGEO databaseWGCNA, SVM-RFEAUC>0.7
20208Zhao et alRheumatologyElectronic health recordsLogistic regression, LASSO penalised and multimodal automated phenotypingAUC: 0.93
Assisted classification and stratification202314Zhang et alJournal of Digital ImagingCT images3D convolutional neural networkValidation set AUC: 0.91, 0.80, 0.96 Test set AUC: 0.94, 0.82, 0.93
202015Castro-Zunti et alComput Med Imaging GraphCT and ageRF, K-NNAUC>0.9
202216Bressem et alRadiologyMRIDeep learningInflammatory changes: AUC: 0.94
Structural changes: AUC: 0.89
202217Lin et alRheumatologyMRIAttention U-netAUC: 0.92
202018Tenório et alInt J Comput Assist Radiol SurgMRI-based radiomics biomarkersMann-Whitney UAUC>0.8
Monitoring the condition202320Baek et alArthritis Res ThermSASSS and clinical dataArtificial neural network, generalised linear modelANN RMSE: 2.83
GLM RMSE: 2.99
201921Gossec et alArthritis Care Res (Hoboken)Wearable activity trackersMulticlass BayesianPositive predictive value 91%
Evaluate the efficacy of the drug202022Lee et alSci RepThe baseline demographic and laboratory dataANN,logistic regression, support vector machine, random forest and XGBoost modelsANN model AUC: 0.783, logistic regression, support vector machine, random forest and XGBoost models (AUC: 0.719, 0.699, 0.761 and 0.713)
202223Wang et alJAMA Netw OpenThe baseline demographic, laboratory data and medication historyLogistic regression, linear discriminant analysis, SVM, GBM and RFAUC>0.7
202124Barata et alJMIR Med InformElectronic medical recordsJoint models/
Prediction of comorbidities202227Zhang et alFront GenetGEO databaseLASSO, RF, XGBoost and SVMSVM: AS AUC 0.7, LBMD AUC: 0.76
202029Navarini et alRheumatol TherClinical dataSVM, RF and KNNAUC values for the ML algorithms were: 0.70 for SVM, 0.73 for RF and 0.64 for KNN
  • ANN, artificial neural network; AUC, area under curve; 3D, three-dimensional; GBM, gradient boosting machine; GEO, Gene Expression Omnibus; GLM, generalised linear model; K-NN, K-nearest neighbour; LBMD, low bone density; PPV, positive predictive value; RF, random forest; RMSE, root mean square error; SVM-RFE, support vector machine recursive feature elimination; WGCNA, Weighted Gene Co-expression Network Analysis; XGBoost, extreme gradient boosting.