Machine learning algorithms that predict the risk of prostate cancer based on metabolic syndrome and sociodemographic characteristics: a prospective cohort study

Table 3 Performance of models after excluding MetS and its components from models

Models	C-index	IBS	D-index
Cox	0.857 (0.855鈥�0.858)	0.015	7.94 (4.90鈥�12.90)
Random survival forest	0.840 (0.839鈥�0.842)	0.016	6.13 (3.78鈥�9.95)
Survival support vector machine	0.862 (0.860鈥�0.864)	-	8.65 (5.32鈥�14.10)
Survival trees	0.836 (0.834鈥�0.837)	0.016	6.53 (3.97鈥�10.70)
Gradient boosting	0.836 (0.834鈥�0.837)	0.015	7.21 (4.34-12.00)
Extra survival trees	0.830 (0.828鈥�0.831)	0.016	5.92 (3.64鈥�9.62)

Note: C-index was estimated based on 100 bootstrapped data samples. Integrated brier score (IBS) applies for models that can estimate a survival function. Thus, it is impossible to estimate the integrated brier score for Survival Support Vector Machine

ISSN: 1471-2458