Development of a deep learning radiomics model combining lumbar CT, multi-sequence MRI, and clinical data to predict high-risk cage subsidence after lumbar fusion: a retrospective multicenter study

Table 2 Comparison of performance metrics

Dataset	Prediction Method	Threshold	ACC	AUC	Sensitivity	Specificity	NPV	PPV	F1
val	Combined Model	0.52	0.79	0.83	0.73	0.80	0.90	0.55	0.63
Test	Combined Model	0.52	0.93	0.93	0.86	0.96	0.96	0.86	0.86
val	Surgeon	–	0.62	0.76	0.27	0.74	0.76	0.25	0.26
test	Surgeon	–	0.53	0.53	0.43	0.57	0.76	0.23	0.30

ISSN: 1475-925X