Conversations about AI and how it will revolutionise the way the NHS delivers care are everywhere these days. Its computational power, speed of data processing, image recognition and ability to automate actions and workflows at large-scale are promised to revolutionise the NHS. From improving staff productivity, speeding up diagnosis, preventing avoidable admissions and reducing pressures on the workforce - AI has the power to revolutionise healthcare.
When examining how we can use AI in healthcare, our team is constantly assessing what will change the game and what is just hype. The stress test is whether it can truly drive exponential gains in care outcomes and / or clinical productivity whilst also maintaining patient safety. In this blog, I’ll go through some of the key considerations regarding the use of AI in primary care, so that we can tackle the challenges that will bring the most impactful benefits.
Building trust and proving value
The reason for the, often invisible, success of AI across other industries is that trust in it has been built over time. As AI is introduced into healthcare, we must recognise the need for this trust building process.
It is predicted that general practice will be one of the most affected workforce groups with the growth of AI in the NHS. There is a lot of potential to help primary care find efficiencies and create more capacity amidst a strained and depleted workforce, but we must ensure we do this in a responsible manner by building trust in the systems and proving value along the way.
Part of this trust will come from regulation - developed through close collaboration between clinicians, software engineers, data scientists and product designers. Another part of this trust will come from how we use AI to complement the highly skilled healthcare professionals in the NHS - assisting them more and more over time whilst keeping them in the driving seat.
Being assistive whilst keeping staff in the driving seat
At Accurx, we believe that utilising AI to assist primary care staff will be incredibly powerful. By automating tasks and suggesting workflows we can enable those talented individuals to use their skills and knowledge to help patients, without becoming overwhelmed and burnt out with the volume of tasks they have to perform. But with this use, it will be important to keep staff firmly in the driving seat.
An analogy that I often use when talking about this is comparing assistive cars vs driverless cars. We need to keep the driver in the driving seat, but give them parking sensors, lane departure warning, cruise control and satellite navigation, to make their driving safer, easier and more productive. It’s not about replacing the driver - like with a driverless car - but recommending the best course of action and being assistive to make the journey as stress-free as possible.
In a GP surgery, this could look like a patient completing an online consultation form saying their back has been hurting for a week, and AI being used to detect key words so it can automatically suggest to the triaging staff member three potential actions to take. One could be sending a questionnaire to gather more information about the back pain, another could be sending a link to the patient for them to book an appointment directly with a physiotherapist, and another could be sending the patient a SMS message with back pain exercise videos. This use of AI for keyword detection and automation could be surfaced and completed within one click, keeping the skilled and trained staff member in the driving seat, but saving time and improving productivity.
Ensuring safety and managing risk
When thinking about the use of AI, it is crucial to ensure that it brings tangible benefits to both healthcare professionals and patients. At the moment, there is some hype around chatbots and symptom checkers for diagnosing and triaging patients in primary care. However, these tools often either have too low a threshold, leading to unnecessary referrals to 111 or A&E, or too high a threshold, resulting in missed diagnoses and patient harm. Given the current pressures on the NHS, trained GP staff should remain integral to triaging decisions. As AI technology evolves and matures, automated triage systems can be integrated to support these decisions, but careful implementation and oversight are necessary to maintain patient safety, as well as building trust over time.
Rethinking the fundamental model of care
Models of care in the NHS have barely changed since its inception in 1948. The introduction of AI would mean overhauling them so they are set up for better outcomes. Just like digitally-enabled care involves more than just digitising existing processes and models, AI-enabled care should not just apply AI on top of existing models. For it to realise the massive productivity gains it promises, a wholesale examination of the patient pathway and staff ways of working needs to be undertaken.
Continue asking questions
I’m excited about a future NHS that fully takes advantage of innovative technology like AI. But as I said at the start, I believe we must continue to ask the question of where AI can make the most difference, so we can understand what will change the game vs what is just hype, and how we build trust in AI over time.
At Accurx, we’re always looking at approaches that will improve our products and their impact on driving positive outcomes for patients and NHS staff. When thinking about AI, we’ll be asking how it can best support those goals.