A popular open-source Spaced repetition memory system. The primary developer is {Peter Bienstman}, a Belgian photonics scientist.
Mnemosyne has been collecting anonymized user data since 2006, but I can find surprisingly little work done with that data. An email to Peter confirms that there have been no serious analyses. Odd.
One significant challenge with this data, like with SuperMemo’s, is that users have few or no overlapping prompts. So for a given prompt, there’s probably just one sample for accuracy at a given repetition.