When people hear “AI validation,” it often sounds abstract. But validation is actually very simple in principle. Andrew explains how they validated an algorithm in Geneva as a lab-developed test —similar to how histology labs validate IHC or PD-L1 assays. -You define your requirements upfront. -You decide what metrics matter -You agree on those metrics with clinical experts. -And then you test the tool in real workflow conditions to see if it actually meets them. For example, run it in routine p