Human respiratory physiology analysis made easier

Before I studied medicine I was a programmer, and I suppose I never quite stopped thinking like one. During my PhD I spent time in England working on advanced analysis of respiratory-physiology data, learning from several people who knew the field far better than I did. What struck me, and frankly frustrated me, was how much of the work was still done by hand. Every manual step was another place for an error to creep in, quietly, without anyone noticing.

I also began to realise how these analyses are passed on. Many groups run their work on a setup handed down through several generations of researchers: numbers go in, analyses come out, and to varying degrees no one still in the room can say exactly what happens in between. The setup gets used largely because the group has published with it before. That is an understandable reason, but it was not one I could work from. I had no such inheritance of my own, so I built my analysis from scratch and, given my background, coded all of it into an automated setup.

It was never meant to be a product. But colleagues grew curious, asked to borrow the code and try it on their own data, and I shared it, because there was no reason not to. A couple of years later, at a conference, I met Nicolle Domnik from Canada, who told me she was already using it herself. Out of that conversation came a different idea: not code passed quietly between friends, but a genuinely public, open-source library that anyone could read, and where everyone gets the same updates.

That conversation became a collaboration. Over the following years we worked on validating and developing the library further, and Yannick Molgat-Seon became a central contributor and part of the group. With help from junior researchers Annie Liu and Nathaniel Yuen, we carried out a larger validation study, presented at the European Respiratory Society International Congress 2022. We are now preparing the full paper for publication, with the support of the Danish Respiratory Society (Dansk Lungemedicinsk Selskab); we will link it here when it appears.

None of us has been paid for any of this. RespMech is a con amore project, built in our spare time because we think it matters. The analyses are demanding, and anyone who takes them on deserves a tool that carries them over the hard part rather than another black box to trust on faith. That is the tool we set out to build, and the reason we keep it in the open.