Skip to main content
. 2019 Mar 7;11(3):328. doi: 10.3390/cancers11030328

Table 3.

Analytical performance of machine learning with random optimization in the testing set.

Performance Parameter ML-RO-0 ML-RO-4 DSS Model a
F-measure b 0.696 0.677 0.698
Accuracy 0.853 0.838 0.860
Area under the curve (AUC) 0.822 0.813 0.815
(+)LR (95% CI) 9.1 (4.3–20.8) 8.5 (3.9–19.6) 8.6 (4.2–18.0)
(−)LR (95% CI) 0.4 (0.3–0.6) 0.4 (0.3–0.6) 0.4 (0.2–0.5)
HR (95% CI) 10.7 (4.6–24.8) 10.3 (4.5–23.7) 10.9 (4.5–26.6)

LR: Likelihood ratio; C.I.: Confidence interval; HR: Hazard ratio; a Analytical performance was evaluated after categorization 0/1 based on risk estimate achieved by both predictors; b F-measure represents a harmonic mean of precision [(P) positive predictive value in machine learning] and recall [(R) sensitivity in machine learning] and is calculated as: 2PR/(P+R).