It's not just a hammers and nails problem
One of the key reasons that biotech ML projects get stuck at the proof-of-concept stage is when they solve a problem that the bench team has no way to act on. This is usually a symptom of a hammers and nails problem: The stakeholders know what their problems are (the nails) but they don’t know all the potential solutions (the hammers). When they come to you with a problem, it’s usually one that they think is solvable rather than the one that’s causing the most pain. So you may never hear about the most painful problems, even if you secretly have the perfect hammer.
So I’ll start this post by reminding you, whenever a stakeholder comes to you with a problem, you should ask as many (potentially naive) questions as you can get away with, to try to find the nail that they actually care about. But there’s an even deeper issue here, specifically in biotech, that’s the real motivation for this post: Biotech data teams often don’t even know all the hammers at their disposal.
For a data team, the hammers that you have available are primarily determined by the data that you have access to, or that you could collect with a reasonable amount of effort. This is true whether you’re displaying that one key number at the right time and place, or building a cutting-edge neural network to predict things beyond human ability.
But it’s not your data team collecting the data, it’s your colleagues in the lab. They’re the ones who know both what’s possible and how trustworthy it is.
At this point the carpentry analogies start to get a bit tortured, but here’s the summary: You need to work with your colleagues in the wet lab to understand both the nails they care about AND the data they can collect for you to build the hammers for those nails.
So that’s twice as many (potentially naive) questions to ask.
This newsletter is brought to you by Merelogic Consulting. We’ll help you turn your ML proof-of-concepts into tangible impact.