In the last decade, presentations to UK emergency hospital departments have increased by 42%. Long-Term Conditions (LTCs) including diabetes, heart diseases, and clinical depression, constitute 61% of these admissions. People who manage their LTCs through continuous treatment, pharmacological or otherwise, have typically 38% fewer emergency admissions. Management of LTCs may benefit from personalised interventions to prevent admission.
AI-Enabled Health Coaching
The NHS Long-term Plan mandates better LTC ‘self-management through personalised support and the use of technology. In this context, Artificial Intelligence (AI) technologies combined with health coaching interventions can help facilitate the identification of individuals in the greater need of support requiring personalised care and improve self-management. Kaplan describes AI as “a system’s ability to correctly interpret external data, to learn from such data, and to use the learnings to achieve goals and tasks through flexible adaptation.”
I worked with HN Company who deliver Proactive tele-Health Coaching (PHC), a personalised care programme to reduce hospitalisation. They utilise Artificial Intelligence (AI) risk prediction to identify people at risk of deteriorating health. Individuals are then reviewed and selected by clinicians to receive support from a health coach in managing their conditions. Over 6-9 months, the coach guides the person through a programme designed to educate and motivate people to achieve their health improvement goals.
Participants develop knowledge, skills, and confidence in managing their well-being, health and care. This approach instils self-confidence and awareness of the condition. Additionally, they received help with identifying their reasons for contacting health services and addressing triggers for urgent access. These reasons may include social isolation and navigating the healthcare system.
Can AI really help?
I supported HN in evaluating the impact through self-reported activation levels and health outcomes data. We asked two questions first, what is the association between the AI-enabled PHC intervention and general, physical, and mental health outcomes? second, did the AI-enhanced PHC intervention change self-activation in every person at high risk of hospital admission?
The evaluation included 288 patients and was nested within a randomised control study of 3,000 people across nine UK geographical areas. These people were on average 75 years of age, had at least one long term condition and were at high risk of emergency admission. We collected measures before they took part in the health coaching, 6 months later and finally after 12 months.
People were better at managing their health
People were better at self-managing their health condition and experienced improved physical health outcomes after receiving the PHC intervention.
There was a 33% improvement per patient in the average self-management activation level score and a 9% change in physical outcomes in the average score from 40.48 to 44.17. The self-management activation level played a role in predicting general, physical, and mental health outcomes.
Population health information including, sex, age, and socioeconomic status indicators such as living environment, housing, employment, were crucial factors at some but not all of the activation levels. For instance, women had a higher activation level and better physical health outcomes compared to men.
Future for AI in primary care?
There is growing evidence about the impact of AI-supported interventions on health outcomes. Back when I was working on this in 2018-2019 it was still novel.
In my thinking, local commissioners could use the evaluation results as a basis to improve and extend the evaluation. It may help them understand the relationship between AI-enabled intervention, patient activation and health outcomes. But, it is crucial to look beyond the statistics and focus on experiences of the health professionals and people.
And finally, expand the AI-enabled intervention to be used in primary care. For example, to identify the impact of demographics, particular illnesses, and service utilisation.
Mariana Wieske is an Associate Partner for Innovex Partners and an advisor for various organisations. She is interested in understanding how digital technologies can support patient’s health outcomes.