Observatory

I think of the work we’re doing as an “observational study laboratory.” So we’ve collapsed that long phrase into one word: Observatory. That works particularly nicely since an observatory is a place built for observing, and that is exactly what observational research does.

What are we doing? We are building tools for observational studies. We will be talking about them here and making them available as they become more polished. The basic tools will be free. There is no point in proprietary tools when everything we are building rests on freely available vocabularies and data. At some point we will have to find a way to generate some revenue, of course — it costs money to make free things. But there are many business models that work. GitHub’s approach, where public work is free and private team work is paid, is one possibility.

Our main tenets are these:

1. Code sets are the fundamental building block. They represent the clinical constructs that an observational study is built from.
2. Complete, but not captive. Our tools should work with whatever approach someone already uses to create cohorts.
3. Use AI where it works reliably. Take advantage of what it does well while minimizing its limitations.

Our home is observational.study — about as literal as we could make it. And yes, “.study” really is a top-level domain, like “.com” and “.org”.

If you want to see where we’re starting, our first post is Common Data Models Answer Half the Question. It also appeared on our jigsaw.io blog, and it marked our move to this site. Jigsaw is still an active project, but it is a closed system and can’t readily be shared in its current form. A lot of the tooling we will be sharing is based on what we learned from building Jigsaw, using it, and updating it for over a decade.