Table 3.

Mean training and validation scores for the best performing clinical- and metabolic tumour volume (MTV)–based logistic regression models

Model	Selected features	Hyperparameters	Mean train score (95% CI)	Mean validation score (95% CI)
Logistic regression – clinical	Cancer stage 1, cancer stage 4, age	C: 10, penalty: l2, Solver: newton-cg	0.74 ± 0.004	0.74 ± 0.02
Logistic regression – MTV (1.5 × mean liver SUV	MTV	C: 1e-07, penalty: l2, Solver: liblinear	0.63 ± 0.02	0.63 ± 0.10
Logistic regression – MTV (4.0 SUV)	MTV	C: 1e-07, penalty: l2, Solver: liblinear	0.62 ± 0.02	0.61 ± 0.10
Logistic regression – clinical and MTV (1.5 × mean liver SUV)	Cancer stage 1, Cancer stage 4, Age, MTV	C: 1, penalty: l2, Solver: saga	0.75 ± 0.004	0.74 ± 0.02

l2, Ridge regression penalty; liblinear, a library for large linear classification