TABLE 3.

Performance of the survival prediction models based on Clinical Characteristics and 24 HLA-Class I, II Superalleles implemented using various regression techniques.

Method	Feature Set-1		Feature Set-2
All Features
	HR	P-value	HR	P-value
LASSO	3.17	3.50E-11	4.05	4.01E-13
RIDGE	3.01	1.76E-10	3.80	2.30E-12
RF	3.09	2.87E-11	3.77	8.15E-12
DT	2.25	6.93E-07	2.00	5.29E-05
Clinical features without tumor status
LASSO	3.50	3.93E-13	3.46	1.54E-11
RIDGE	3.49	3.93E-13	2.97	2.89E-09
RF	3.74	3.01E-14	2.96	8.23E-10
DT	2.15	2.24E-06	1.83	3.12E-04
Clinical features without tumor stage
LASSO	2.80	9.96E-10	3.51	1.32E-11
RIDGE	2.43	4.68E-08	3.55	7.56E-12
RF	2.81	2.05E-10	3.18	2.14E-10
DT	2.50	1.64E-08	2.76	2.38E-09
Clinical features without tumor stage and tumor status
LASSO	2.40	4.41E-08	3.11	5.60E-10
RIDGE	2.40	4.41E-08	2.57	5.81E-08
RF	2.99	9.37E-12	2.59	1.55E-08
DT	2.54	1.06E-08	2.65	7.37E-09

#HR: Hazard Ratio; RF: Random Forest; DT: Decision Tree; Feature Set-1: Clinical and demographic features, Feature Set-2: Clinical, demographic and 24 HLA superalleles features.