OpRESTORE is a national NHS service that supports UK veterans with complex physical health problems linked to their military service. Veterans referred to OpRESTORE often need care from many different specialists, including surgeons, pain teams, rehabilitation, and mental health services. Currently, decisions about which service is most appropriate are made by a multidisciplinary team (MDT) of clinicians. While effective, this process can be slow, resource-intensive, and sometimes difficult for patients to navigate.

This study will develop and test a new digital "navigator" tool that uses artificial intelligence (AI) to support these referral decisions. The aim is to see whether the tool can safely and accurately match veterans to the right care pathway, while reducing delays and improving patient experience.

The project will be carried out in several stages:

• Reviewing past OpRESTORE records to design the AI model.
• Testing the tool alongside the MDT ("shadow testing") to check whether its recommendations match the clinical decisions.
• Running a case-control study to compare outcomes between patients referred using AI support and those referred by the MDT alone.
• Creating and testing a structured self-referral form to make it easier for veterans to access care directly.

The main outcome will be whether the AI tool makes the same referral decisions as the MDT. Other outcomes include patient satisfaction, quality of life, time taken to reach the right service, and overall costs.

The study will recruit veterans aged 18 or older who are referred to OpRESTORE with a physical health need. It will run for two years. If successful, this approach could free up clinician time, shorten waits for treatment, and improve veterans' health and wellbeing, while laying the foundations for wider use of AI-supported navigation across the NHS.

BACKGROUND

Introduction OpRESTORE was established in 2016 to provide specialist multidisciplinary care for UK military veterans with service-related physical health needs. Over the past seven years, more than 1,500 patients have been referred into the service, reflecting both the high burden of complex conditions in this group and the unmet need for coordinated care pathways. OpRESTORE has demonstrated clear benefits: patients report high satisfaction, almost 60% of cases are redirected to a more appropriate pathway, and statistically significant improvements are seen in EQ-5D-5L outcomes . However, referral rates are rising by more than 30% annually threatening to exceed clinical capacity. The challenge is not a lack of expertise but one of scalability. To continue delivering high-quality, timely, and equitable care, a digital evolution of the OpRESTORE pathway is required.

Why OpRESTORE was needed OpRESTORE arose in response to systemic inefficiencies in the NHS patient pathways. Over the last decade years in good health have declined, with 15-20% of the population affected by musculoskeletal pathology, a leading cause of disability and economic inactivity and the bulk of OpRESTORE's workload. Primary care is increasingly overstretched: for the first time in two decades, patient dissatisfaction with GP services exceeds satisfaction. Access is particularly limited in deprived communities, contributing to widening health inequalities. For veterans, these challenges are compounded by fragmented medical care following transition out of the Armed Forces, difficulties navigating civilian healthcare systems, and the psychosocial impact of leaving service life. Scepticism over civilian providers ability to understanding the nuances of military-related conditions also deters veterans from accessing NHS care.

Referral pathways are a major bottleneck. 7-9% of GP appointments result in a referral, yet 18% of patients require four or more consultations before being referred. Crucially, 21% of patients fall into the "referrals black hole", where appointments are cancelled, misdirected, or lost to follow-up. These delays prolong the already long referral journey: in 2022, 25% of people in England were waiting for an appointment, test, or intervention, and 59% of patients did not receive specialist care within the 18-week target. The net result is a system in which GPs shoulder a rising workload, specialists face inappropriate or delayed referrals, and patients experience inequitable, protracted and fragmented care. Veterans, who often present with multi-morbidity, complex physical needs and often psychosocial challenges, are especially vulnerable to these failures.

Healthcare navigation as a solution

Healthcare navigation offers a proven solution to fragmented systems. The evidence base is extensive: one meta-analysis and 13 systematic reviews were consulted for this project, in addition to relevant individual publications. Collectively, these studies demonstrate that navigation programmes consistently improve access to care and enhance patient outcomes. Key findings include:

• Increased screening uptake,
• Improved attendance to appointments, follow-up adherence and treatment compliance,
• Decreased time to diagnosis and treatment,
• Reduced unplanned care including emergency visits and hospitalisations ((8,16,17) and shorter hospitalisations when they do occur,
• Improved patient satisfaction and outcomes, particularly for chronic disease and underserved groups,
• Reduced mortality.

Navigation has also been shown to deliver measurable cost savings. Examples include reductions in cancer care spending of $1,676 per patient per month and savings of $17,780 annually for patients in the NavSTAR substance misuse programmes. Systematic reviews consistently conclude that patient navigation shortens hospital stays and decreases unplanned care, both of which are major cost drivers. Financial modelling studies similarly predict significant per-patient savings in transitional care and cancer pathways.

The OpRESTORE patient navigator has proven some of these benefits. Over half of patients are re-routed to a different pathway than originally anticipated, ensuring more appropriate use of scarce specialist services. Patient Reported Outcome Measures (PROM; EQ-5D-5L) demonstrate measurable improvement in quality of life, and patient surveys report high levels of satisfaction with the service. In short, OpRESTORE has proven the concept of navigation for veterans in the UK context.

The emerging challenge: workload and scalability Despite these successes, OpRESTORE now faces an inflection point. With referral volumes growing by more than 30% annually, the workload threatens to outstrip available staff. Continuing to expand purely through human capacity is unsustainable. What is required is a digital augmentation of the pathway, working smarter, not harder. Some elements of navigation, particularly triage of straightforward cases, are amenable to automation. However, this must be achieved without sacrificing the patient-centred and "human touch" elements that underpin OpRESTORE's success. An AI-solution could streamline workflow, decrease workload and free up human resource to add value through direct patient encounters rather than bureaucratic and administrative processes.

A further dimension is the potential for self-referral models. Across Europe, healthcare systems have increasingly turned to restrictive gatekeeping, limiting direct specialist access in the hope of controlling costs amid ageing populations and rising demand. Yet such approaches risk delaying care for those who genuinely need it. AI-enhanced patient navigation offers a more balanced alternative: enabling efficient self-referral for patients who warrant specialist care, while still ensuring intelligent gatekeeping through AI-facilitated triage.

Evidence gaps and the role of AI While patient navigation is well established in cancer care and some chronic disease contexts, there is little published research on AI-enhanced healthcare navigation, particularly in the UK and across multiple specialties. Current published literature mostly consists of study protocols or future projects and opinion pieces. Reviews highlight the potential of AI to support triage, personalise patient education, improve care coordination, and optimise resource use. However, robust, real-world evaluations are lacking.

Our team is addressing this gap through a systematic review of AI in healthcare navigation (PROSPERO 2025 CRD420251019208). This will inform the design and evaluation of a digital, AI-enhanced version of the OpRESTORE pathway.

Not only are the digitalisation and intelligent automation of OpRESTORE critical for its continued growth and success, but it is also uniquely positioned to serve as a test case for pioneering AI-patient navigation in UK veterans. It is safe, with an established multidisciplinary clinical safety net. It has almost nine years of patient-level data across >1,500 cases. The case mix offers both complexity (multi-morbidity, service-related pathology) and simplicity (nearly 50% remain on the same pathway, and there is high homogeneity with musculoskeletal referrals making up 74% of cases). This balance makes OpRESTORE both a rigorous and feasible setting to pilot AI-enabled navigation.

Policy alignment Importantly, digitising OpRESTORE aligns with UK health system priorities. The Delivery Plan for Recovering Access to Primary Care (2023-25) highlights the importance of self-referral and digital transformation to relieve GP burden and improve access to healthcare. The NHS Long-Term Plan and the Darzi Review (2024) both emphasise accelerating digitalisation, harnessing AI, and shifting towards technology-enabled efficiency in patient navigation. The Tony Blair Institute (2024) has underscored AI, specifically for triage and navigation, as a critical driver for sustainable healthcare reform.

Conclusion OpRESTORE has proven the value of patient navigation for veterans in the UK but faces an unsustainable rise in demand. Digitisation, informed by robust research and AI-enhanced tools, offers a safe, scalable pathway forward. By combining evidence-based navigation with national policy priorities, the proposed research will generate insights with relevance not only for veterans, but for wider NHS and international healthcare systems.

RATIONALE FOR CURRENT STUDY This project aims to evaluate whether an AI-enhanced patient navigation tool can replicate MDT decision-making while decreasing clinician workload. This study is designed to inform the staged implementation of a digital navigation pathway, including self-referral, clinician-verified algorithmic triage, and, ultimately, selective automation for appropriate cases.

Research Question:

Can an AI-enhanced digital patient navigator safely and accurately replicate MDT decisions in the OpRESTORE pathway, while decreasing clinician workload? Can it improve system efficiency, and patient-reported experiences and outcomes?

Hypothesis:

A digital, AI-enhanced patient navigation tool will be able to guide patients to appropriate care pathways with the same reliability as MDT decision-making, while requiring less manual input. Additionally, structured self-referral will capture richer clinical data than traditional GP-led referrals, leading to improved patient experience and engagement.

2. STUDY OBJECTIVES

Primary objective:

To determine whether a digital AI-enhanced navigation algorithm is equivalent to MDT decision-making in directing patients to appropriate care pathways.

Secondary objective:

• To evaluate patient experience (PREMs) and PROMs across different referral pathways.

• To identify patient cohorts suitable for automation (defined as 80-95% initial algorithmic accuracy).

• To assess time to MDT discussion and referral issuance.

• To conduct a cost analysis comparing standard and digital workflows.

• To benchmark MDT decisions against NICE guideline-based recommendations.

  3. STUDY DESIGN

Type of study: Digital Health Intervention Development and Evaluation - A mixed-methods digital health and AI algorithm development study, involving retrospective analysis of clinical records, prospective observational validation, and a future interventional case-control trial of an AI-based healthcare navigation tool.

Duration: 2 years

Subjects: Veterans accessing the OpRESTORE patient navigation pathway.

3.1. STUDY OUTCOME MEASURES

Primary endpoint:

The concordance between algorithm-generated care-pathway recommendations and MDT decisions.

Secondary Endpoints:

1. The accuracy of referrals, as determined by independent experts, being equal to or superior to that of the MDT.
2. The increase in EQ5D-5L scores (baseline to exit) of patients under the algorithm-supported pathway being equal to or greater than those under standard pathway.
3. Patient-reported experience measures (PREMs) assessed post-referral (e.g. overall satisfaction, ease of use, accessibility) of patients under the algorithm-supported pathway being equal to or greater than those under the standard pathway.
4. Time from referral to MDT decision and referral issuance being faster in the algorithm-supported pathway.
5. Cost per referral episode being lower in the AI-supported pathways.
6. Proportion of algorithm-assisted referrals providing strong support for a treatment pathway and not requiring further input from the MDT, with a target of 40%
7. The suitability of this platform for a subgroup of veteran amputees.

Inclusion Criteria:

• Meets criteria for referral to OpRESTORE service
• Age 18 years or older.
• Capacity to consent.
• Have a physical health need (e.g. not purely mental health, or seeking social care advice)

Exclusion Criteria:

• Referrals processed outside standard MDT workflow.
• Patients lacking capacity to consent.
• Prisoners.
• Acute presentation best managed by emergency services and not appropriate for OpRESTORE.