Table 6.

Performance metrics across predictive models: observation period ⩽1 year after index date (n = 4501; outcomes = 1646).

Metrics	Lasso logistic regression	Naive Bayes	Gradient-boosting machine	Random forest	Neural network	Random	All positive	All negative
Accuracy, %	66.22	60.89	67.64	66.76	59.47	50.67	63.20	36.80
Matthews correlation coefficient, %	18.98	14.96	23.97	21.20	11.59	−0.44	0.00	0.00
Area under curve, %	68.43	59.96	69.21	68.12	59.04	50.61	50.00	50.00

NOTE: outcome was progression to later line of therapy after ⩾60 days from the end of a previous one or a stem cell transplantation procedure.