Table 4.

Summary of clinical approaches.

Category	All variants	GIAB benchmark only	Nonactionable only	GIAB benchmark + nonactionable only
Risk of reporting false positive	Higher	Lower	Higher	Lower
Risk of adverse impact on patient care	Higher	Higher	Lower	Lower
Not eligible—no passing model	6	6	6	6
Not eligible—outside benchmark region AND actionable	N/A	N/A	N/A	4
Not eligible—actionable	N/A	N/A	48	44
Not eligible—outside benchmark region	N/A	88	N/A	84
Eligible—predicted true	240	164	216	141
Eligible—predicted false	60	48	36	27
Confirmation order rate	21.57%	46.41%	29.41%	53.92%

This table summarizes the prospective results for all variants (n = 306) under different clinical approaches. The methods are organized from highest risk of reporting a false positive to lowest, where reporting actionable variants without confirmation is considered highest risk. Approaches that allow the models to be applied to any variant interpretation (specifically primary or actionable) have a higher risk for adverse impact on patient care. Approaches that allow for variants from any genomic region (specifically outside Genome in a Bottle [GIAB] benchmark regions) have a higher risk of reporting a false positive. Variants that are classified as “not eligible” either did not have a validated model or require confirmation test due to the approach. Confirmation order rate is the percentage of variants that are either not eligible or predicted false, indicating that a confirmation test would be ordered for that variant prior to reporting. The results from our clinical approach (nonactionable only) are emphasized.