Table 2.

Table showing the minimally curated and highly curated model names and a description of the training dataset.

	Model	Description
Mixed-Curation models	Models 1–8	2% (56 scans) random samples of mixed-curation training dataset
	56-scan	Model 1, best performing 2% random sample of mixed-curation training dataset
	142-scan	5% random sample of mixed-curation training dataset
	284-scan	10% random sample of mixed-curation training dataset
	710-scan	25% random sample of mixed-curation training dataset
	1420-scan	50% random sample of mixed-curation training dataset
	2130-scan	75% random sample of mixed-curation training dataset
	2840-scan	100% of mixed-curation training dataset
Highly curated models	56-scan	Random sample of highly curated training dataset to match 56-scan minimally curated model
	142-scan	Random sample of highly curated training dataset to match 142-scan minimally curated model
	244-scan	Full highly curated training dataset