The challenge

Across the health system, demand is rising faster than services can respond. An ageing population, increasing multi‑morbidity, and growing complexity of need are placing sustained pressure on acute services, with care often delivered reactively rather than preventatively. 

Many patients with long‑term conditions experience fragmented pathways, limited follow‑up, and gaps in coordination between health, social care and community services. As a result, needs frequently go unrecognised until they escalate into crisis, driving avoidable hospital admissions and poorer outcomes. 

Delivering more proactive, preventative care requires integrated neighbourhood teams able to identify need early, coordinate across services, and intervene before escalation. However, this is constrained by workforce shortages, limited capacity for proactive outreach, and the challenge of translating data into consistent, coordinated action at scale. 

Persistent inequalities further compound the challenge, with populations experiencing deprivation and poorer health often least likely to access timely, preventative support, and most likely to present late with complex need. 

These challenges are particularly evident in the South West, where an older population and higher prevalence of multi‑morbidity increase the complexity and volume of need, placing additional strain on already stretched services. 

System pressures are further compounded in rural and coastal communities, where dispersed populations, access barriers and isolation mean patients are more likely to become “invisible” between contacts, with unmet need emerging later and more acutely. Fragmented services and workforce constraints make delivering proactive, coordinated neighbourhood care even more difficult in these settings. 

The solution

BRAVE AI is an artificial intelligence tool, designed and developed by Bering Ltd, that uses machine learning to score a patient’s health complexity and predict their risk of sudden medical crisis, especially in more vulnerable ageing populations. This helps clinical teams deliver more timely and proactive interventions to support personalised care and reduce deterioration and unplanned admissions. 

BRAVE AI is helping clinical teams to organise care differently, helping teams move from prediction to action. It enables earlier intervention, better coordination, and more consistent proactive care, laying the foundation for longer-term reductions in avoidable demand and unplanned admissions. 

The tool was originally piloted in Somerset, where an innovator- and site-led evaluation in care homes showed promising results: a 60% drop in emergency attendances, 35% fewer falls, and 8.7% fewer ambulance call-outs.   

Expanding on this success, NHS England South West supported the wider roll out of BRAVE AI as part of its Digital Neighbourhoods Programme and through additional support from the Health Technology Adoption and Acceleration Fund (HTAAF). BRAVE AI is now implemented in 47 primary care networks across England.  

Our work in partnership

The last decade charts a journey of successful adoption and spread of BRAVE AI in the South West, supported by Health Innovation South West, helping patients and health and social care providers focus on prevention rather than cure.   

Having worked with Bering Ltd since 2014, Health Innovation South West’s support for BRAVE AI has included:  

• Grant funding for prototype development, product testing, and building partnerships with local systems and early adopters  

• Evaluation support to demonstrate the impact of implementation  

• Support through HTAAF for further roll-out and evaluation  

• An SBRI phase 3 grant including HISW support for real-world evaluation  

Evaluating to inform national spread and adoption  

In 2025-26, we carried out a real-world evaluation of BRAVE AI impact and implementation across three Primary Care Networks in Somerset, Herefordshire and Suffolk, with Small Business Research Initiative (SBRI) Phase 3 funding which we helped to secure. 

Across these sites, BRAVE AI has been implemented within proactive care pathways for a range of cohorts, including care home residents, patients with frailty, dementia, those who are housebound, and those who have been identified through local area and deprivation-based health analysis.  

The mixed‑methods evaluation examined how the tool was used in practice over a six‑month period, focusing on whether improved identification and prioritisation of patients translated into proactive, coordinated care. 

Our evaluation and learning experts worked closely with participating sites to define evaluation priorities and approaches, develop practical tools to support cohort management within multidisciplinary teams, and establish consistent methods for data collection across the pilot. 

The evaluation combined quantitative data on activity and care processes with in‑depth insight to understand how BRAVE AI is used in practice, exploring how and why it works, and what it contributes to proactive care, decision‑making and collaborative working to provide a balanced assessment of impact and to inform future spread and sustainability. 

Patient and public involvement was also a key component. We co-designed accessible materials with Bering and patient representatives, conducted interviews with members of the public on their perceptions of BRAVE AI and proactive care, and fed back key insights to support programme communication strategies.  

Our impact so far

1. Previous Somerset Care home pilot showed 60% fewer ED attendances, 35% fewer resident falls, and 8.7% fewer ambulance call-outs.  
2. HTAAF 2025 findings suggested minor reductions in elective and non-elective admissions, and a 30% improvement in integrated neighbourhood working.
3. The SBRI 2025/26 evaluation has shown stronger MDT prioritisation and collaborative working, improved patient reassurance and proactive support, more structured operational delivery of Population Health Management. With further impact described below

Demonstrating impact via our real world evaluation 2025-26: Initial findings   

Our real-world evaluation carried out in 2025-26 has demonstrated that BRAVE AI enables proactive, population-based care when it is embedded in routine multi-disciplinary team workflows and linked to action. Its value lies not in predicting risk alone, but in helping teams identify and act on patients with unmet or actionable need.  

Across the three sites in Somerset, Herefordshire and Suffolk over the six-month implementation period, 635 patients were identified and reviewed, and more than 460 patients received proactive intervention or follow-up.  

A consistent pattern was observed across all sites: once patients were accepted into pathways, most progressed to proactive review and intervention. 

By cohort:   

• In dementia, 33 patients were identified, 31 (94%) of which received a proactive review and follow-up 

• Among care homes and housebound patients, 109 patients were identified, 64 (59%) of which were reviewed in multi-disciplinary teams (MDT) and almost all acted on 

• In frailty, 79 patients were identified, 75 (95%) of which received action 

• In locality population (SWH), 414 patients were identified, 290 (70%) of which received outreach, follow-up, or intervention to varying degrees. 

Across sites, BRAVE AI-supported implementation was associated with a substantial shift from reactive and inconsistent activity towards more structured, proactive and coordinated care. 

Staff findings suggest that BRAVE AI improved the structure, consistency and confidence of multidisciplinary prioritisation and proactive care delivery. Importantly, staff highlighted that impact depended on having the workforce, MDT structures and coordination capacity to act on identified need. 

The strongest early effects were seen in identification of unmet need, increased proactive intervention, stronger MDT coordination, improved care planning, and better patient experience. Longer-term outcome impact is likely to depend on sustained operational capacity, pathway maturity, and the ability of neighbourhood systems to translate insight into coordinated action. 

What’s next?

The evaluation will be published in full in Summer 2026 alongside an independent Health Economic evaluation conducted by SWART Evaluation Ltd, and learnings and recommendations will be used by Bering Ltd and system partners across the country to scale spread and adoption of BRAVE AI in proactive, neighbourhood-based Population Health Management. 

In the South West, we will be building on our work with BRAVE AI as we assess neighbourhood assets and models of care in our region – where rurality and coastal geography, workforce challenges, and high prevalence of multi-morbidity and complex needs, combine to demand new models of proactive, community-centred care.  

What’s been valuable about this programme is the opportunity to learn in real time. Each site approached implementation slightly differently, which has helped us understand how local context and workflow shape adoption in practice.  

As an independent evaluation, our focus has been on capturing that variation and providing a balanced, evidence-based account of what’s working, what’s been more complex, and what can reasonably be concluded from the six-month pilot.” 

– Nic Ferreira, Interim Head of Neighbourhood Health, Health Innovation South West