AI guided clinical coaching in urgent and emergency care

Share

Background

There is an extraordinary amount of pressure on urgent emergency care services both in London and across England, and the situation is getting worse.

To be able to manage the pressure and achieve high-quality results – while making sure that the system is sustainable long term – we need a more proactive and preventative approach. UCLPartners, North East London Integrated Care System, and Health Navigator are currently delivering the AI for Urgent and Emergency Care programme, which aims to support patients before they reach crisis points, helping them live healthier lives and easing pressure on A&E services.

The programme uses routinely collected local hospital data to create bespoke machine learning algorithms that predict high-intensity users (people who access the service regularly) of unplanned care – people who often seek emergency or urgent healthcare services without appointments – up to six months in advance. Advanced AI screening technology is used to identify patients who are more likely to need unplanned emergency care, and they are offered support via targeted, phone-based clinical coaching with trained healthcare professionals (clinical coaches). Clinical coaches support patients to self-manage their condition, and reduce their chances of unplanned hospital visits, admission, and extended hospital inpatient spells.

THIS Institute is carrying out an evaluation of the project, along with LCP Health Analytics, to gather evidence that will support its wider adoption and help shape future policies for effective proactive care.

Approach

We will be working with LCP Health Analytics to independently and objectively assess the impact, effectiveness and delivery of the AI for urgent and emergency care programme. We will do this by designing and delivering an effective Real World Evidence evaluation approach, which involves using data collected from real-life settings, rather than controlled clinical trials. Jointly with UCLPartners, we will produce data for regular ongoing learning, and insights to improve the impact and effectiveness of the programme. The work will be carried out across four workstreams.

In Workstream 1, we’ll evaluate whether the AI-guided clinical coaching intervention reduces unplanned hospital admissions, use of healthcare resources and clinical outcomes such as the incidence of new acute and chronic conditions and mortality among patients who have been identified as at high risk of needing unplanned care. This will be done through a matched cohort study – a study comparing two groups of people, designed to mimic the setup of a clinical trial. This target trial emulation approach (a framework for designing and analysing observational studies that estimates the causal effect of interventions) uses electronic health record data.
In Workstream 2, we will assess how well the coaching intervention is adopted, accepted, and put into practice. We will analyse data from people who were offered clinical coaching, and interview patients, carers, and healthcare professionals. We’ll use sub-group analysis to explore the differences in implementation and impact across different groups and equality dimensions (like sex, gender, age, ethnicity, disability).
In Workstream 3, we will evaluate whether clinical coaching improves patients’ knowledge, skills and confidence to manage their own health, along with their quality of life. We’ll carry out an electronic survey of patients who received the coaching intervention using the thiscovery.org platform.
In Workstream 4 we will conduct a health economic evaluation, aligned with HM Treasury Green Book guidelines, to understand the cost-benefit and budget-impact at system and national level of AI-guided clinical coaching.

Project team