News and Insights
Ways NHS uses AI to tackle healthcare challenges
June 11, 2024
The four key ways the NHS is harnessing artificial intelligence to address challenges right now
As any casual observer of the UK’s general election campaigns will have seen, the NHS faces a multitude of serious pressures, causing near-record high levels of patient dissatisfaction, extensive staff burnout and stubbornly high waiting lists for planned care. In recent years, rarely a day has passed without the consequences of these competing pressures leading morning news bulletins and monopolising the newspaper front pages.
Islands of experimentation
It is against this backdrop that we all see the need for faster care and better patient experience scraping up against the reality of major workforce shortages and creaking physical infrastructure, aggravated by the fact that our clinical IT systems aren’t yet as good at talking to each other as those used by comparable health systems across the globe.
This is where some care providers are finding creative uses for AI-based tools. These islands of experimentation are still at a relatively early stage, but all share the same goals – to improve outcomes for patients, help clinicians and the public at large. These tools fall into four broad use cases.
Waiting list management
The AI tools currently being used to efficiently prioritise those in need of acute or elective care harness this concept in increasingly sophisticated ways. One prominent example is a data analytics platform that triages patients on waiting lists for planned care. It uses data already kept on clinical IT systems across different care sites to deliver risk scores for each patient, reflecting measures based on its determinations of the risk of complications, risk of mortality and change in overall mortality, to name a few.
In the North West of England, the system has been used in conjunction with a coaching service to identify high-risk patients awaiting surgery and to offer them specialised support to get them in the best condition possible ahead of their surgery date.
An NHS England assessment of the first 125,000 patients managed through this pilot programme found a two-thirds reduction in the number of high-risk patients requiring a visit to an intensive care unit.
Diagnostics and imaging
One of the multitude of factors behind the UK’s 7.6 million-strong waiting list for planned care and medical procedures is a shortage of diagnostic capacity. As of January, there were just over 1.5 million patients waiting for a key diagnostic test.
That is why it is encouraging that this is an area where AI is having a demonstrable impact on the timeliness and quality of treatment. Even those advocating for more sceptical analysis of AI’s problem-solving abilities in healthcare acknowledge its potential in image recognition, particularly in identifying cancers and other irregularities in X-rays, CT scans and MRI imaging.
When properly applied as a support tool for specialists the AI-based tools can help to reduce referral and treatment times for patients, while providing extra sets of eyes to spot signs of ill health from scans. These systems can provide multi-disciplinary teams with real-time information while working remotely, meaning that the right support can be offered across several sites quickly.
In recognition of its potential in this area, last autumn the government created a £21m AI Diagnostic Fund to provide tools created by academic research consortiums and specialist tech vendors to speed up lung cancer diagnosis. This funding pot has been shared among 64 NHS trusts across the country.
Elsewhere, tens of thousands of stroke patients across England have been seeing quicker treatment thanks to a system that provides instant interpretation of brain scans, dramatically reducing the time taken for those who have suffered a stroke to get the right treatment.
This e-Stroke system, developed by a company spun out of the University of Oxford’s Nuffield Department of Clinical Neurosciences, uses specialised algorithms to interpret brain scans and inform clinical decisions in real time.
Documentation and administration
NHS England has been proactive in supporting provider organisations to adopt tech-enabled solutions to seize back thousands of hours of clinical work that would otherwise be monopolised by typing notes into electronic (and paper-based) patient records.
Platforms available to NHS organisations integrate with a provider’s electronic patient record, allowing clinical and nursing staff to dictate notes directly into relevant records. The tool can also draw information from a patient’s record to quickly generate patient follow-up letters based on pre-programmed templates. In addition to the clinicians’ time being saved, provider organisations and their commissioners have also benefited from financial savings in outsourced transcription costs.
These generative AI tools automatically create summaries of patient conversations based on recorded consultations. A crucial feature of this type of platform is the increasing sophistication with which it can accurately identify symptoms in order to help rule out certain conditions.
In an environment where organisations are suffering chronic shortages of both administrative and clinical staff and weeks-long turnaround times for appointment and referral letters to reach the patient, this type of innovation has been instrumental helping to make care providers more efficient in ways that make the most difference to those receiving care.
Predictive analytics and prevention
Advanced healthcare systems understand the importance of effective prevention on both their populations’ quality of life but also the impact it has on healthcare spending. Systems that identify vulnerable individuals and at-risk groups can do a great deal to ensure that small interventions earlier on can prevent costly emergency care or the need for on-going chronic condition management.
The NHS in the south west of England has been pioneering the use of a machine learning-based risk assessment tool to reduce preventable hospital admissions. The system uses algorithms to identify patterns in patients’ GP records to assess the likelihood of unplanned emergency ambulance call outs and emergency department visits over the next 12 months.
An early pilot programme initiated by NHS Somerset analysed data from over 500 care home residents and resulted in a 35% reduction in emergency department visits.
The system, which is being rolled out to 30 groups of GP practices across the south west throughout 2024, can then inform preventative care efforts to those frail and vulnerable patients deemed most at risk. This includes inviting individual residents to take part in a wider assessment of their health and wellbeing, which then informs a personalised care and support plan drawing in help from local nurses, pharmacists, therapists, health coaches and social prescribers.
NHS England has said that if the current roll-out is successful, then the platform will be deployed across all GP practices across the south west.
What comes next?
These are snapshots of how AI is being deployed tentatively across the health service. There are of course other pockets of AI-driven innovation in areas as diverse as drug development, talking therapies, and spotting early signs of sepsis.
Clearly more work needs to be done to show feasibility, safety and efficacy in a wealth of the many heralded healthcare applications for AI. Behind the scenes, the government, health service and the regulatory bodies that surround the health and care system have embarked on some striking initiatives to build this evidence base while building the regulatory guard rails.
One such piece of work is the Medicines and Healthcare products Regulatory Agency (MHRA) AI-Airlock scheme, set to launch this month (April). The AI-Airlock provides developers with a kind of accelerated access to patients under the supervision and oversight of regulators. This way, certain types of AI technology to be used within an NHS setting before gaining full regulatory approval. The key purpose of this is to generate evidence on how effective a system is while prioritising patient safety.
As an organisation so rich in data, the NHS represents an incredibly valuable resource for ‘training’ AI systems and the kind of academic and scientific research needed to bring genuinely transformative tools to healthcare. Of course, a great deal of caution and healthy scepticism is needed to ensure that only evidence-based solutions that operate on a legally sound basis make it out of the ‘sandbox’ and into the hands of clinicians, patients and commissioners.
But as things stand, the health service is showing signs of sensible, pragmatic thinking in its approach to an unfolding shift in the way that we use technology to improve health outcomes across the board.
Explore the healthcare communications services FINN Partners offer to help your organisation navigate the evolving landscape of AI use cases within the NHS.
TAGS: Health
POSTED BY: Nick Renaud‑Komiya