Proactive frailty care for care home and housebound patients
How one PCN used Brave AI to identify and act on unmet need in its most vulnerable patients, reducing falls by 89%, cutting medication burden, and delivering near-universal MDT review.
Context
A PCN with two distinct high-need populations
This case study is drawn from an independent evaluation of Brave AI's SBRI Phase 3 programme, conducted by Health Innovation South West. It focuses on a Suffolk PCN that used Brave AI to support proactive care across two of its most vulnerable patient groups: care home residents and housebound patients.
Prior to implementation, MDT involvement across these cohorts stood at just 6.2%. Proactive, personalised care planning was virtually absent. Patients were being managed reactively, seen when they presented with problems, rather than reviewed in advance. The PCN set out to change that.
Approach
Using Brave AI across the full risk spectrum
The PCN used Brave AI differently across its two cohorts, reflecting different population sizes and clinical needs.
Care home cohort
Brave Scores were used to generate a RAG-based MDT review across the full risk spectrum, not just the highest-risk patients. This enabled the team to identify patients with moderate or rising risk scores who had actionable unmet need, rather than focusing exclusively on those already known to be in crisis.
Housebound cohort
Given the smaller cohort size, RAG banding was not applied. Instead, Brave AI was used to prioritise the timing of reviews, helping the team decide who to see first and when, based on changes in score and complexity over time.
In both cases, the tool was used to interrogate unmet need and identify service gaps.
Results
From 6% to 98% MDT involvement
Of the 109 patients identified, 64 (59%) were accepted into the MDT pathway following clinical validation. Of these, almost all received active review and intervention. MDT involvement among accepted patients rose from a baseline of 6.2% to 98.4%. A near-total transformation in how this population was managed.
The MDT process generated multi-domain interventions across every accepted patient. Common actions included:
Medication optimisation
Review and rationalisation of polypharmacy
Falls assessment
Structured falls risk review and prevention planning
Anticipatory care planning
ReSPECT and personalised care and support plans
Functional review
Support coordination and referral to community services
Clinical impact
Measurable improvements in patient safety
The programme demonstrated measurable improvements in two clinically significant areas for frailty patients.
Falls reduction
2.3 → 0.2
Mean recorded falls per patient. An 89% reduction associated with proactive frailty assessment and intervention.
Medication burden
44.9 → 39.4
Mean medication items per patient. Rationalisation through structured MDT review and pharmacy involvement.
Both outcomes are particularly significant for frailty populations. Falls are a leading cause of hospitalisation and loss of independence in older adults. Polypharmacy is independently associated with adverse drug events, falls, and frailty progression. Addressing both through proactive MDT review, rather than waiting for a crisis, is central to the Brave AI model.
Patient experience
Feeling heard, supported, and prepared
Across the programme, patient and carer feedback was consistently positive. The evaluation found that the most immediate and consistent benefit was not rapid clinical improvement, but increased reassurance, better coordination across services, and greater preparedness, particularly for patients and families navigating complex or deteriorating conditions.
"Helped identify those patients who had limited interaction with health care but who would benefit from a timely MDT assessment and intervention.""— Geriatrician
"The patients and family have been very grateful of this interaction.""— Care Coordinator
"Completing the Brave AI project has brought the MDT together to look at care of patients and ensure that all aspects of required input have been given. Also it has encouraged discussion across the MDT on patients that would not have been discussed. This is leading to preventative input.""— Allied Health Professional
The evaluation found that 46 of 50 patients across all sites reported that being proactively contacted helped them feel more confident about their care. 44 reported that the conversation reassured them or helped them feel more supported in managing their health.
Implementation learning
What made it work
The evaluation identified several factors that drove the PCN's strong outcomes.
MDT capacity came first
The strongest predictor of impact was whether the team had the MDT capacity, care coordination infrastructure, and workforce skill mix to act on what they found. Identification without action changes nothing.
Clinical validation was non-negotiable
The team never used Brave AI outputs at face value. Every patient list was reviewed, contextualised, and refined using clinical judgement before action was taken. The tool was a starting point for enquiry, not a determinant of care.
Focus on actionable need, not highest risk
Rather than defaulting to the highest Brave Scores, the team used the tool to identify patients whose needs were both significant and actionable. This included moderate-risk patients, rising-risk patients, and those with visible service gaps.
Impact grew over time
The first one to three months involved clarifying MDT roles, refining the target cohort, and building confidence in interpreting scores. Staff consistently reported that perceived value increased as the tool became embedded in routine practice.
Source
This case study is drawn from an independent process evaluation of the Brave AI SBRI Phase 3 programme: From Prediction to Action: Evaluating BraveAI within Proactive Neighbourhood Models, conducted by Health Innovation South West (Libby Smith, Nic Ferreira, Richard Blackwell, May 2026). Site and patient details have been anonymised in accordance with the evaluation's anonymisation framework.