Just a guess: Ratio of positive results out of previously unknown cases gives a good estimate of the total expected unknown cases. Precision depends on selection bias, and statistics has tools to estimate and/or improve it.