Explore more realistic test creation

[sumiyaabdi]

Could use LSC + short windows for more very detailed diagnostics

• Example shiny for Alzheimers: https://data.darwin-eu.org/p4-c1-021-diagnostics/
• Diagnostics code: https://github.com/darwin-eu-studies/P4-C1-021-Diagnostics

From adam:

• We could run some detailed diagnostics and then create synthetic test data from that diagnostic data.
• see if we can run large scale characteristics with lots of different time windows
• Use small time windows (maybe 30 days) before and after index. Look maybe 1 year before index and 5 years after index, something like that. Also characterize death in time windows as well.