TABLE 3.

Performance across probability thresholds for the High-risk PARS classifier (logistic regression). Metrics for precision, recall, and F1 refer to the positive class (High-risk PARS = 1).

Threshold	Accuracy	AUC	Precision (class 1)	Recall (class 1)	F1 (class 1)
0.30	0.908	0.931	0.795	0.808	0.802
0.40	0.902	0.931	0.822	0.733	0.775
0.50	0.900	0.931	0.886	0.650	0.750
0.60	0.896	0.931	0.912	0.608	0.730

Bold values indicate the selected threshold (0.30), which achieved the highest F1 score while preserving recall.