Development and validation of radiology-clinical statistical and machine learning model for stroke-associated pneumonia after first intracerebral haemorrhage

Table 2 Multiple model evaluation indexes for different model

Model evaluation index	Support vector machine		Random forest		Logistic regression		ISAN
Model evaluation index	Training group	Testing group	Training group	Testing group	Training group	Testing group	Total group
AUC	0.812	0.674	0.684	0.652	0.796	0.746	0.688
95% CI	0.783–0.842	0.602–0.747	0.650–0.718	0.580–0.724	0.762–0.830	0.677–0.815	——
Sensitivity	0.793	0.395	0.471	0.463	0.387	0.395	0.149
Specificity	0.832	0.874	0.897	0.841	0.940	0.974	0.904
Positive predictive value	0.573	0.593	0.566	0.439	0.644	0.593	0.304
Negative predictive value	0.934	0.756	0.856	0.854	0.844	0.756	0.790
F1 score	0.665	0.474	0.514	0.450	0.483	0.474	0.200
Accuracy	0.823	0.722	0.803	0.761	0.818	0.722	0.527

ISSN: 1471-2466