I just heard about this on the radio and found this article discussing it. It is being used in Israel and has been used in the past for other disorders.

The advantage that the method has over just simply grouping samples is that if there is a positive result, the method can determine who it is, even if there are a few people with positive results. The example given used 48 samples for 384 people, which would allow for one test for every 8 people. I don’t know the exact way that it is done, but each person is part of 6 of the pooled samples. The combination of positive results from the 48 samples is a key that tells who was responsible.

I have seen elsewhere that the method used is based on something called the Reed-Solomon error correction code. I have a rough idea of how error correction codes work, but do not see how this relates to test pools. If any of you see the connection, please tell what it is.