The manufacturer of a diagnostic test (a machine learning model) for a very rare disease that affects about 0.001% of the general population (a "positive" condition) claims that the diagnostic test is 95% accurate in identifying people who don't have the disease (a "negative" condition) when tested against a true random sample drawn from the general population. The manufacturer wishes to sell this machine learning model to hospitals at a cost of $1.49 million per seat license.

Is this diagnostic test valuable or not? Why or why not? If you were a hospital administrator, would you buy it?

Group of answer choices

Yes, because a model that identifies 95% of people who don't have the disease is a very valuable test. I would certainly buy it!

No, it is a useless test, because the random sample is highly imbalanced and therefore accuracy is not a good measure. I would not buy it!