science

The study did not present the dogs with samples from "years before" and did not have an accuracy of 98%. It had a sensitivity (percent of positive samples called positive) of 70% for one dog and 80% for the other. Their specificity (percent of negative samples called negative) was 90% and 98%.

Since there were 60 positive samples and 40 negative samples, their accuracies were 82% and 91%, respectively.

Not 98%. And with samples from patients currently diagnosed (or controls).