Lost in translation.
The request reached me through a friend: a company needed an "infant model." We both assumed it meant a 3D model, the kind of asset I'd been building for games for years.
The company was AveXis, a biotech firm developing a treatment for spinal muscular atrophy (SMA), a genetic disease that progressively weakens the muscles of infants and young children. AveXis was later acquired by Novartis.
Changing course mid-sentence.
On the call were AveXis's researchers, a room full of PhDs and MDs. As they described their work, it became clear they didn't need an asset at all.
They had recorded infants with SMA using a Microsoft Kinect, a motion-sensing camera built for video games. What they needed to know was whether those recordings could be used to objectively measure movement and track how the disease was progressing.
I changed course in the middle of the conversation and scoped the real problem on the spot.
They asked for an asset. They needed a diagnostic pipeline.
The game default was the wrong answer.
I'd spent nearly two years working on Kinect Star Wars, so I knew something about Kinect that most people outside games wouldn't.
Kinect's body tracking is built for games, and games want clean avatars. So the system deliberately filters out "extraneous" movement and snaps players toward tidy, predictable poses.
For diagnosing SMA, that extraneous movement is the signal. The small, irregular movements a game would throw away are exactly what clinicians need to see. The standard game-industry approach would have discarded the most important data, and you'd only know that if you'd built with it.
A yes, a method, and a plan.
In roughly 30 hours of consultation, I:
- Confirmed the recordings were usable, and that the data the researchers needed was in them.
- Devised a method to extract the movement data they needed from the existing recordings.
- Delivered a phased roadmap for building a diagnostic tool, structured so each phase gave them something they could make decisions with.
Along the way, I explained game technology to medical researchers in terms they could act on. That kind of translation between specialists is a large part of how I work with game teams too.