Healthcare Experiment Planning

Use this skill to design experiments that test a memory-native system, not preprocessing tricks or capabilities an EHR already provides.

Workflow

1. Name the decision-worthy problem first.

   • State who cares: board, CQO, CMIO, patient-safety lead, researcher, or engineer.
   • State what existing tools fail to do: not what is merely inconvenient.
   • Reject ideas that only prove chart lookup, SQL filtering, or ordinary EHR reporting.

2. Write the primary question in one sentence.

   • Prefer: "Can the full stack assemble the right cross-time evidence packet from noisy memory better than narrower baselines?"
   • Avoid: "Can the system retrieve rows we already identified with patient/time filters?"