AI for Urgent and Emergency Care in North East London: Using Predictive Analytics and Patient Partnership in Integrated Care

Background: Urgent and emergency care (UEC) systems in England face increasing pressure from rising demand, workforce shortages, and persistent health inequalities. To address these challenges, NHS England funded a 3-year National Demonstrator programme called AI for UEC with NHS North East London, in partnership with Health Navigator and UCLPartners, as a Real World Evidence study on predicting and preventing avoidable hospital use through proactive, person-centred support. Covering 2.5 million residents, this is one of the largest real-world studies of AI for integrated care, aligning directly with the ICIC26 theme “Enhanced Value-Based on Prevention and Technology.”

 

Approach: This ongoing programme (2024 - 2027) builds on 10 years of robust evidence from multiple randomised controlled trials (RCTs) and peer-reviewed studies demonstrating that predictive analytics, combined with personalised clinical coaching, improves patient outcomes and reduces emergency activity.

Using routinely collected healthcare data, we trained local machine-learning models (adjusted for equity) to identify individuals at rising risk of unplanned admission.

Those identified by the model and screened as eligible by clinicians are offered voluntary enrolment into a proactive clinical coaching intervention. The intervention focuses on activation, motivation, and self-management of long-term conditions, complementing existing clinical and community pathways.

 

An independent evaluation has been established to assess patient outcomes, health economic impact/system activity savings, plus implementability and patient acceptance of such interventions. This mixed methods evaluation led by UCLPartners and delivered by two evaluation partners: THIS Institute (University of Cambridge) and LCP (health economic consultancy).

Patient and Public Involvement and Engagement (PPIE) has been planned for the duration. The underlying approach has been co-designed over a decade with patients, and clinicians.

Results:

At the time of abstract submission the service has been running for approximately 11 months with >3,000 patients onboarded to the Clinical Coaching service.

Interim operational findings using propensity score matching demonstrate statistically significant reductions in unplanned hospital activity (bed days) as well as mortality impact. This provisional data exceeds impact seen from original RCT data including 25% reduction bed day reduction in intervention cohort.

At the time of the ICIC26 Conference we will have interim data from the independent evaluation to present. The evaluation protocol includes propensity score matching (1:4 matching ratio) with an intervention cohort size of 5,500 at time of the conference. We will provide sub-analysis on protected characteristics including ethnicity, and deprivation.

 

Implications: The North East London experience demonstrates that AI-enabled preventive care is feasible, effective, and scalable:

1. Feasibility: Existing NHS data and governance frameworks are sufficient to deploy predictive solutions safely and with public trust.
2. Effectiveness: The approach delivers tangible patient and system benefits without exacerbating inequalities, and may actively mitigate them by targeting support to those most at risk.
3. Scalability: The design is transferable within the UK and internationally, offering a practical blueprint for health systems facing similar UEC and population-health pressures.

Together, these insights show that technology, when co-produced with patients and leveraging the energy of patient themselves, can deliver true value-based prevention -strengthening integrated care systems and improving outcomes at scale.