The Slow Registry Habits

Each chapter of the book teaches one habit. The habits follow the life of a registry, and the chapters can be read in order or on their own.

Part One: Before anyone enrolls

1. Decide who the data are for. Name the users of the data and let the most important one decide.
2. Start with the analysis plan. Decide the first product, a paper, a trial-ready cohort, or a partnership, and plan the data backward from it.
3. Look at who is already in the space. Amplify an existing effort before building a new one.
4. Decide what to capture before you go looking. Know what the record holds, and ask for the rest.
5. Put a person between the form and the participant. A coordinator participants look forward to hearing from.
6. Know who belongs on the data team. Match the expertise to the question.
7. Settle the protocol, dictionary, and consent before enrollment opens. Governance cannot be added later with AI.

Part Two: Running the registry

8. Verify the diagnosis before you count it. Evidence fitted to the purpose, read by a qualified person.
9. Keep records someone else could read. Every value and every cohort has a trail.
10. Track every sample and every link. Every identifier leads back to one consented person.
11. Hold the weekly meeting. Everyone who touches the data, together, every week.
12. Go and find out why the data are missing. Every blank has a cause.
13. Stay in touch on a schedule. Retention is a calendar of human contact.

Part Three: The community and the field

14. Be part of the community, not a website. Show up, hold office hours, report back.
15. Keep the researchers close. Researchers use registries they know.
16. Have the conversations only you can have. FDA, industry, and the room AI cannot sit in.

Part Four: AI on top of all this

17. Read what the machine wrote. The summary is generated; the reading is yours.
18. Build so that a machine can use it. The habits above are what make AI useful.