Key Takeaways

Dear reader,

If you take one idea from this book, take this one: a registry is made by people, and the tools come after.

Walk through one registry year with me. Before anyone enrolls, you decide who the data are for, look at who is already in the space, and settle the protocol, the dictionary, and the consent. You assign the identifier that will follow each person, and you give families an approved place to tell you what you did not think to ask. You name the coordinator, even if the coordinator is you, and you name the people who will work with the data every week and the advisors you will call.

Then enrollment opens, and the year settles into a schedule. Every week the team meets and walks the cohort: who enrolled, who is overdue, whose diagnosis still needs a qualified reader, and what looks wrong. Someone tags every document on the day it arrives and follows every sample back to one consented person. When data that should exist are missing, someone goes and finds out why, and often recovers them. The newsletter goes out, office hours are held, and a family that asked a hard question gets a call back.

Through the same year you are part of a community and a field. You show up at the family conference whether or not anyone enrolls. You find the researchers working on your mechanism in a neighboring disease and tell them what living with the condition is like. You ask the consortium what it would need to see before you collect the measure you hope to qualify. Every data request goes through the committee, the IRB, and the agreement, and each one teaches you something about your data.

At the end of that year you have a data set a sponsor, a researcher, or a regulator could use. You can say how many people are enrolled, how many are confirmed and by whom, what is missing and why, and where every number came from.

That is also the data set an AI tool can work with. I use AI every day, and I expect to use it more. The habits in this book are what make it useful on registry data, and none of them can be delegated to it. A tool can draft, summarize, and code. It cannot decide what your registry is for, earn a family's trust, read a genetic report with the judgment of a credentialed person, notice that data which should exist are absent, or sit in the room with a regulator.

None of this is fast, and all of it can be learned. Many of the people who do it best did not train for it. They are parents, caregivers, and people living with a condition who started because no one else would, and who kept going one habit at a time. Start small, take a beat before you build, ask the people who have done it, and put a person at the counter.

Thank you for the work you do for your community. I would like to hear how it goes.

XO, Danielle