As we're starting to analyze data for the SRDeviceUsageReport we've found a small number of duplicates. Out extraction strategy for SensorKit in general is to keep the date of the last sample retrieved and start the next SRFetchRequest at essentially 1 millisecond after that time so we don't get the same sample again and conceivably don't miss anything in between.
On 3 participants looking at iPhone data, with about 5000 samples reviewed, we found about 1 % of samples were seen more than once - spanning an extraction cycle. More troublesome we found about 0.5% were ALMOST equivalent, but overlapping. So since these reports come in 15-minute segments, one might be 2022-05-12 12:00:00 - 12:15:00, but the next one might come in at 2022-05-12 11:59:00 - 12:14:00. And sometimes they have totally different data!
Curious if anyone else has experienced this or has any advice on how to avoid/make sense of the data in that situation.