NPfIT failed for other reasons mainly
Including being not very good.
This was partly because it was run by managers who had limited understanding of IT and less of the business; and specified by clueless people from clueless firms.
Talking about medical re odd (what is one of those BTW) rather than automation is tending to be an index of being behind; and a major failure mode is seeing computers as devices with which people can be programmed.
GPs inventented this stuff, put it into operation in their Practices, in several cases actually wrote it, and have now lost control of it to the elements above.
I did play golf, once, in 1974 IIRC which by coincidence was about the time The London Hospital implemented a computer system that was useful on the wards. I have a picture somewhere of the 10 MByte hard drive being craned from its lorry into the building.
The example of the bankers may not be entirely good, but consider the idea that if you want to help someone clever, knowledgeable and hard working to do something they are good at a bit better faster or cheaper you should probably be sitting near them watching and listening, and building software in conversations which occasionally include "make this script work fast and reliably"
You'd better also reflect that much of medicine is nonlinear.