fbpx
Connect with us

AI

New horizons: Why AI could be the life-saver the NHS has dreamed of

Published

on

Odgers Berndtson’s technology and healthcare partners sit down with Tara Donnelly, experienced CEO and board director, and former chief digital officer at NHSX and NHS England, where she was responsible for a range of technology, digital and data programmes across the NHS.

Tara now works independently as Founder of Digital Care. They discuss the enormous benefits the NHS could reap by using AI, as well as some of the challenges the health service might face in its implementation.  

The NHS has numerous AI projects in play.

It is already used in the majority of stroke cases, helping clinicians reduce the time to deliver care by an hour, and tripling the number of recoveries.

A cloud-based AI technology deployed within the NHS this year, is helping specialists plan radiotherapy treatments ‘two and a half times faster’ than if they were working alone.

While AI teledermatology, which can predict cancerous lesions with 99.7 per cent accuracy, is currently being rolled out.

Imaging, including mammography and eye scans, digital pathology, head and neck radiotherapy, and data analysis are among the areas currently being transformed by AI.

But according to Tara Donnelly, this is just the tip of the iceberg.

What benefits will AI bring to the NHS? 

“I’d love to see the power of AI harnessed with a focus on giving clinicians the gift of time.

“Doctors are likely to be perceived by patients as more compassionate and showing greater empathy when they are freed from the slavery of constant documentation – through smart ambient documentation recognising speech with accuracy.

“Being liberated from that allows eye contact, active listening and the opportunity to really focus on the patient, not having to capture the record.”

“Routine use of remote monitoring devices, combined with AI health analysis and recommendations will mean far few health ‘crises’, affording a range of staff – from paramedics to GPs – significantly more time.

“They’ll be able to work from home and conduct more visits to patient homes or to community health hubs.”

This isn’t just wishful thinking – a vision of this future already exists.

‘Virtual wards’ allow patients – including those with acute and long-term conditions – to recover in their own homes.

For example, using virtual wards, Airedale and Imperial NHS Trusts, have halved the number of readmissions for long-term conditions, such as COPD, a progressive lung disease, and heart failure.

Combined in the future with large language models and AI, the impact to both patients and the NHS could be revolutionary.

“AI, used in conjunction with remote monitoring, could review vital sign information and insights into patients with multiple health conditions, helping clinicians manage co-morbidities.

“Using this data, the AI can act as a clinical co-pilot to the team, prioritising those most in need.

“This will enable clinical staff to focus on those most at risk and support them with personalised, proactive care – it’s going to offer ‘wrap around’ recommendations for multiple conditions to keep people well, and at home for longer.”

A less talked about benefit will be job satisfaction.

With more time, less burnout, and unencumbered from the drudgery of manual searching and documentation, clinicians will be able to focus on using clinical judgment and delivering personalised care.

Their jobs, and those of many other healthcare workers, will become more interesting. This in turn is likely to have a positive impact on patients; happier staff means more empathetic care.

What challenges will the NHS face in rolling out AI? 

“Data quality is, and will continue to be, a huge issue.

“With home monitoring, part of the joy is people can mobilise freely, but movement can affect the signal and readings being fed in and create false positives.

“Data sets used to train AI need to represent the world’s population if they are going to be used in providing care to diverse populations.

“London is quickly becoming an AI health research hub in this regard – think of the work at Moorefields, UCLH, Imperial, the Royal Free, to name a few – because it has the world’s population, a real advantage.”

This matters – a lack of diversity in data sets means AI risks replicating historical biases and even providing unhelpful and incorrect recommendations to those groups it has not been trained on.

Recent research shows this could lead to unfair outcomes for patients, particularly in detecting disease from blood samples or imaging data.

“Codes of conduct and strengthened regulation are also going to be important – simply because not much exists at present.

“Clinician acceptance, or lack of, may create some resistance to AI adoption in the NHS.

“Most doctors want to understand underlying causes, and so – for excellent reason – would want to understand why something has been flagged by the AI.

“AI models have traditionally been bad at this – too much of a black box – and so if we’re to get clinician acceptance, AI models will need to demonstrate the rationale for their decision making.”

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending stories