Fig. 9: Evaluation of the BOUND model’s performance on the geography-based external validation under completely random missing rates of 10%–70%. | npj Digital Medicine

Fig. 9: Evaluation of the BOUND model’s performance on the geography-based external validation under completely random missing rates of 10%–70%.

From: Interpretable machine learning model for digital lung cancer prescreening in Chinese populations with missing data

Fig. 9: Evaluation of the BOUND model’s performance on the geography-based external validation under completely random missing rates of 10%–70%.

a The intercept of the calibration curve, with values closer to 0 being preferable. b The slope of the calibration curve, where values closer to 1 are optimal. c ROC curves, with the area under the curve representing the AUC. d Decision curves.

Back to article page