A nearly-launched clinical investigation will explore how AI can be used in live point-of-care workflows to support the proactive delivery of guideline-based care and prevent emergency hospital admissions for people with Chronic Obstructive Pulmonary Disease (COPD).
COPD affects around 1.2 million people in the UK and is the second most common cause of emergency hospital admissions.
The annual economic burden of COPD on the NHS is estimated as £1.9 billion with treatment following exacerbation of symptoms accounting for 70 per cent of these costs.
The DYNAMIC-AI clinical investigation is a 12-month feasibility study now underway in consenting patients with COPD resident in NHS Greater Glasgow and Clyde (NHSGGC.
The study uses machine learning based models to identify patients at highest risk of adverse events for review by clinicians.
In the study, a COPD multi-disciplinary team will consider model outputs in real time with the aim of allowing proactive interventions to improve outcomes and reduce emergency care requirements.
The collaboration between Lenus Health and NHSGGC is the first to operationalise predictive AI in routine direct care of chronic conditions.
The project follows extensive co-design efforts with patients and clinicians to develop a technically viable and clinically explainable way of delivering machine learning derived risk scores into existing care pathways.
Dr Chris Carlin, consultant respiratory physician at NHSGGC, who is leading the investigation, said:
“This is an incredibly exciting project.
“It’s the first time we’re bringing together predictive AI insight for COPD into live clinical practice.
“With the ageing population and rising prevalence and complexity of long-term conditions, clinicians are overwhelmed with data that they don’t have the capacity to review.
Moving from rule-based care to machine learning models
Lenus Health’s team of data scientists and engineers have pioneered the development and training of four machine learning models to proactively identify patients with COPD who are at risk of adverse events and provide actionable insights to improve quality of care.
The proprietary AI algorithms are UKCA marked and were trained using almost one million data points from historical electronic health records from a de-identified cohort of more than 55,000 patients with COPD resident in NHSGGC.
Lenus Health uses more than 80 data points to support the delivery or risk scores, significantly more than in traditional rule-based systems, which are known to cause numerous false alarms, leading to clinicians experiencing alarm fatigue at potential risk to patient care.
To address this problem, Lenus Health has invested four years of research in model development focused on fairness and explainability, which enables clinicians to understand and interrogate the model’s scores and ensure they are bioplausible.
Dr Carlin said:
“Rule-based systems are static whereas machine learning is much more robust in the context of routine care.”
Clinical care teams will be given actionable insights from the models to use in multi-disciplinary team (MDT) reviews.
By identifying high risk patients, they can be offered pro-active, preventative care to avoid the COPD symptom flare-ups that currently cause 1 in 8 emergency hospital admissions.
Dr Carlin said:
“One of the key things we hope this will tell us is which patients are at risk of adverse outcome so we can provide anticipatory care.
“We will be able to discuss with patients the level of care they want in advance rather than in an emergency situation which is much more pressured.”
Addressing equality issues
COPD disproportionately affects deprived populations, with the prevalence of COPD in the most deprived 10 per cent of areas in the UK is almost double that of the least deprived 10 per cent.
The project’s work on fairness provides a meaningful scientific contribution to identifying biases held in health data and ensuring models perform appropriately across age groups, gender, deprivation categories and ethnicity.
As a result, it has the potential to improve access to healthcare to people living in socio-economically disadvantaged areas of the country.
Lenus Health CEO, Paul McGinness, said:
“We anticipate that the data science approach, technology infrastructure and wider learnings from this exemplar study will accelerate the application of AI across other long-term conditions to help address growing demands on health systems caused by increasing levels of multi-morbidity.”
