. Author manuscript; available in PMC: 2019 Nov 29.

Published in final edited form as: IEEE Access. 2018 Nov 29;6:77796–77806. doi: 10.1109/ACCESS.2018.2884126

TABLE 3:

Best results AUC and accuracy for diagnostic experiment and risk prediction experiment using conventional features only versus using conventional features and delta features together. Statistically significant results at (p < 0.1) are proceeded by asterisk.

Experiment	Conventional features only			Conventional features and delta features
Experiment	Features set	Best AUC	Best Accuracy (%)	Features set	Best AUC	Best Accuracy (%)
Diagnostic	Definiens	0.833	80.66	Definiens	0.851	82.07
	Rider	0.820	81.13	Rider	0.858	83.96
	Pyradiomics	0.784	79.71	Pyradiomics	0.817	* 83.49
Risk Prediction	Definiens	0.767	75.00	Definiens	0.807	76.41
	Rider	0.773	76.88	Rider	0.822	78.30
	Pyradiomics	0.713	74.52	Pyradiomics	0.731	75.20