Fig. 3: Results of the AI model on the ILD classification.
From: Interstitial lung disease diagnosis and prognosis using an AI system integrating longitudinal data
a AUC comparison between the joint AI model and human readers on the classification of each ILD subtype. The (b) Sensitivity analysis of the joint AI model and human readers’ results. The markers represent the sensitivity of the AI model and human readers on each ILD subtype, and the lines represent the confidence interval of sensitivity. c Specificity analysis of the joint AI model and human readers’ results. The markers represent the specificity of the AI model and human readers on each ILD subtype, and the lines represent the confidence interval of specificity. Each human reader was indicated with different markers. Sensitivity and specificity comparison were calculated via the exact Clopper-Pearson method to compute the 95% confidence interval (CI). In (b, c), data are presented as true sensitivity/specificity + /− 95% CI respectively.
