Healthcare has been an enthusiastic buyer of virtual reality for a decade, with mixed results. The projects that succeeded share a property the failures did not: the thing being trained was genuinely spatial and genuinely difficult to practise any other way.
Three conditions worth testing against
- Practice is dangerous or impossible. Anything performed on a patient qualifies; anything performed on a form does not.
- The skill is spatial or procedural. Sequence, orientation and hand position are hard to convey in video and easy to convey in simulation.
- Repetition matters. If competence comes from doing something forty times, a headset is cheaper than forty supervised attempts.
If a use case fails all three, a well-made web module will beat a VR module on cost, accessibility and the likelihood that anyone completes it.
The content problem is the real problem
The headset is the cheapest part of the programme. The expensive part is authoring clinically accurate scenarios and keeping them current as guidance changes. Any immersive training programme that cannot be updated by clinical educators without engineering support will be out of date within two years.
That constraint should drive the architecture from the beginning. Build the authoring pipeline before the second module, not after the tenth.
Measure competence, not completion
Time spent in a headset is not an outcome. The systems that justify renewal are the ones recording competency per procedure and feeding it somewhere a supervisor already looks — which usually means integration work nobody scoped at the start.