Harnessing AI predictions for smarter prevention: three recommendations for policymakers

The recently published findings from a landmark NHS clinical trial highlight the vast potential of AI-powered preventative care to improve population health outcomes. These findings, stemming from a comprehensive randomised controlled trial spanning over seven years, including eight NHS trusts and enrolling more than 1,800 patients, underline how AI screening tools can combine with nurse-led coaching to make a profound difference to patient care.

The key findings from the report are clear: men at high risk of an adverse health incident over the age of 75 stand to benefit the most from a targeted prevention model. The model uses AI to identify those who will benefit the most and remote coaching to provide truly personalised care. In this group, for 8 males treated, one extra life was saved – a truly extraordinary result. This could be compared to pharmaceutical treatment with statins, which have become synonymous with “heart-attack-and-stroke-preventing,” avoiding one heart attack for every 60 patients treated for five years and one stroke for every 268 patients treated for five years.

As policymakers look to address systemic issues of reactive treatment models and overwhelmed health and social care systems, this emerging evidence signals key opportunities. This paper looks back at the findings and suggests key recommendations for policymakers to implement as we look to build on the potential of AI to improve patient outcomes and lives.

Recommendation 1: Invest in research on AI prediction and risk stratification tools

This trial demonstrated AI's ability to accurately flag high-risk patients for targeted outreach, reducing mortality. But more research is critically needed to validate predictive algorithms across diverse populations and care pathways. Policymakers should look to fund large-scale studies on risk stratification AI to build evidence for national implementation.

Recommendation 2: Develop incentives and reimbursements for preventative AI models

The current system predominantly reimburses treatment activities, not preventative outreach. AI can today support predictive health analytics, patient triage and prevention, and ‘HN Predict’ is already in use with great effect across the UK and Ireland. HN’s model – built by clinicians and undergone robust national trials to improve the allocation of preventative care resources, reducing costs and delivering better patient outcomes – is an important example.

To catalyse adoption of solutions like the trial's AI screening and nursing interventions, policymakers must develop incentives and value-based reimbursements for proven preventative approaches.

Recommendation 3: Support infrastructure for data-driven population health

Effectively implementing preventative population health powered by AI requires sophisticated data infrastructure. Policymakers should work with colleagues in health to invest in:

  • Integrated data systems spanning care settings
  • Robust data security and privacy measures
  • Responsible governance that facilitates access to data while upholding security and ethical standards.

This will lay the groundwork for scalable AI prevention strategies that improve outcomes.

The clinical trial evidence provides a glimpse into the enormous potential of AI-enabled preventative care. With thoughtful policies supporting research, incentives, and infrastructure, the UK can lead the way in leveraging AI predictions to make smarter proactive health a reality. The future of healthcare begins with foresight.

Dr Joachim Werr, Founder and Executive Chair, HN

Steve Laitner, GP and freelance health consultant

The EMJ paper is available to view and download here. A summary outlining the main highlights from the report and includes a patient case study can be downloaded here.