When AI meets human connection: What a behavioural analysis revealed about AI-assisted clinical coaching

Our behavioural analysis of an AI-assisted clinical coaching service highlighted the value of combining AI with trusted, personalised coaching. AI helped identify people who were likely to benefit from support, while trusted coaching conversations helped them build the confidence, skills and motivation to better manage their health.

Much of the conversation around artificial intelligence (AI) in healthcare focuses on what technology can do: identify risk, detect patterns and support decision-making at scale. Equally, we know that some of the most meaningful improvements in an individual’s health happen through human connection: a trusted conversation, personalised support and the confidence that comes from feeling listened to and understood.

To explore this relationship, we recently conducted a behavioural analysis of the clinical coaching service developed by Health Navigator and delivered in partnership with NEL ICB and UCLPartners. The service uses  locally trained and validated AI algorithm to identify patients at high risk of accessing unplanned care and offers personalised clinical coaching to support them in improving their health and reducing future unplanned admissions.

Our analysis explored how the service supports behaviour change, in particular,  patients’ self-management and help-seeking behaviours. Through a review of training materials, focus groups with coaches and interviews with patients, we sought to understand not simply whether the service works, but how it works.

Looking beneath the surface of outcomes

Healthcare innovations are rightly judged by their outcomes. We want to know whether interventions improve experience, reduce demand on services or support better health. Yet outcomes alone rarely tell us how an intervention achieves its impact.

Behavioural analysis helps open that “black box”. By applying established behaviour change frameworks, we identified the active ingredients within coaching conversations and explored the mechanisms through which they influence behaviour.

The analysis found that coaches were already using a broad range of evidence-based behaviour change techniques. These approaches supported reflection, problem-solving and confidence-building, creating a strong foundation for successful self-management.

The power of being identified at the right time

One of the most striking findings was the role of timing. Many patients described engaging with the service at pivotal moments in their lives, such as after a hospital admission, before surgery or during periods of uncertainty about their health. Several felt that coaching arrived at exactly the right time.

This raises an interesting question about the role of AI within the programme. While the technology identifies people at elevated risk, the findings suggest that the service may also reach some patients at a time when they are especially receptive to support and behaviour change. In this way, AI may facilitate opportunities for meaningful intervention.

The relationship that turns opportunity into change

If AI helps identify the opportunity, the coaching relationship helps patients make the most of it. Across patient interviews, trust, rapport and connection emerged as some of the most valued aspects of the service. Patients described feeling understood and supported through conversations tailored to their individual circumstances.

Patients valued having dedicated time and space to reflect on challenges, explore options and build confidence in managing their health. Coaching was frequently described less as receiving advice and more as being supported to find their own solutions. What emerged was a picture of a service that works not by directing patients, but by empowering them.

Building capability, not just knowledge

Another notable finding challenged some common assumptions. Most patients did not report significant gaps in their knowledge of healthcare services. Instead, they described gaining confidence, practical self-management skills and a greater ability to deal with challenges affecting their health and wellbeing.

This reinforces that behaviour change is not always about providing more information. Sometimes it is about helping people feel able to act on what they already know.

From insight to action

The analysis also highlighted clear priorities for strengthening the service: maintain the trusted, person-centred coaching environment that patients value most; equip coaches with practical tools to help patients overcome system-related barriers; and continue building patients’ confidence to advocate for themselves when navigating care.

This is where behavioural analysis can be particularly useful. It does not simply tell us whether an innovation is succeeding. It helps us understand what should be preserved, what can be strengthened and where relatively small changes can create meaningful improvements in people’s experiences and outcomes.

What this means for health innovation

For partners across healthcare, academia and industry, the findings offer an important insight: successful innovation depends on understanding human behaviour as much as technological functionality.

The AI component helps identify individuals who may benefit from support. The coaching model provides personalised, behaviourally informed conversations. The patient-coach relationship creates the trust and engagement needed for those conversations to have impact.

This analysis is a reminder that technology alone is rarely enough to change how people manage their health. The real opportunity lies in combining the capabilities of AI with an understanding of human behaviour and the trusted relationships that help people make meaningful changes to their health.

Author: Orla O’Doherty, Learning and Behaviour Change Lead, ULCPartners