Artificial intelligence (AI) has huge potential to improve healthcare. From earlier diagnosis to more personalised care, it’s already making a difference.
But it’s also under scrutiny. A recent study1 found gender bias in a social care tool used by many councils, raising questions about whether AI could widen health inequalities if not designed and tested carefully.
For innovators developing AI for the NHS, this highlights the challenge: a promising technology alone is not enough. Adoption depends on safety, fairness, and transparency – as well as clear alignment with NHS priorities. Based on our work with AI companies, here are five tips to help you prepare for NHS adoption.
Five tips for preparing your AI innovation for NHS adoption
1. Show real world performance
Some innovators tend to over rely on technical performance. Accuracy matters, but it’s not the full story. A model that performs well in controlled conditions can still struggle in practice. NHS decision-makers will ask how it works across diverse populations and care settings, and how you’ve tested that. Real world experience varies greatly from controlled lab-based models.
Clinical trials or pilot projects are a strong start, but NHS decision makers want evidence of real world impact. Consider the impact to cost-effectiveness, patient experience and health equity. Does it save money for the NHS? Does it improve the quality of care and reduce unnecessary variation in outcomes? It isn’t just performance in ideal conditions. It’s a robust model that stands up to real world usage.
2. Think beyond regulation
Passing regulatory hurdles is essential, but it doesn’t guarantee adoption. Evidence requirements for regulation differ from those needed to support reimbursement or procurement. Innovators who plan for both from the start are much more likely to succeed. Consider whether your innovation needs to be classed as a medical device and, if so, what level. Can you manage with a CE or UKCA mark only and work up to a medical device position?
3. Don’t underestimate the complexity of the NHS
AI tools rarely operate in isolation. They need to integrate with existing pathways, data systems and workflows. Innovators sometimes underestimate how disruptive even small changes can be for clinicians and administrators. Although AI models can be adapted and changed rapidly, humans need time to adjust their working patterns and their confidence in using AI with patients and colleagues. Examine the current pathway that patients follow and consider how your AI will positively disrupt this pathway to benefit patients. This shows you understand how the NHS works now and how it could be improved with your innovation in place.
4. Concentrate on the government’s three shifts for the NHS
The NHS is focused on three priorities: moving care closer to home, preventing illness earlier, and accelerating the shift from analogue to digital. AI that supports these shifts will always get more traction than solutions that don’t clearly align. Consider how you are supporting any or all of these shifts and how this will affect patients and the NHS, both now and in the future.
5. Plan for the implementation journey
Even when the evidence is strong, adoption can stall if patients or the NHS workforce isn’t prepared or pathways aren’t adapted. Innovators who plan for implementation early – including training, support and service redesign – remove one of the biggest barriers to uptake. Health is a complicated industry and one that needs to be measured and careful in its delivery.
How we can help
At Health Innovation Oxford and Thames Valley, we help innovators shape adoption and reimbursement strategies, connect with regulatory experts and align with NHS needs. Working with us means your innovation has the best chance of becoming part of NHS care.
For practical tools and guidance on preparing your innovation for the NHS
Ready to take the next step? Get in touch info@healthinnovationoxford.org
About the author
