Table 4.

Classification of (1) resection vs. non-resection ROIs and (2) seizure-free vs. not seizure-free patients, using random forest and a linear support vector machine.

	Accuracy		Sensitivity	Specificity
	Mean	Confidence interval	Mean	Mean
Resection vs. non-resection ROIs
Random forest	60.34%	59.98–60.71%	61.45%	59.52%
Support vector machine	59.94%	59.67–60.22%	55.07%	64.82%
Seizure-free vs. not seizure-free patients (MEG metrics only)
Random forest	49.03%	47.25–50.82%	49.40%	48.67%
Support vector machine	43.77%	42.08–45.45%	42.60%	44.93%
Seizure-free vs. not seizure-free patients (MEG and clinical metrics)
Random forest	49.74%	48.14–51.35%	50.55%	48.94%
Support vector machine	42.95%	41.54–44.36%	42.12%	43.77%