The healthcare sectors in Canada and the UK are facing similar workforce pressures – can ambient AI provide rapid solutions?

Cindy Fedell, Regional Chief Information Officer, Northwestern Ontario Hospitals in Canada, with her 7+ years experience in the NHS in the North of England, is uniquely placed to look at how AI solutions can help address the pressing workforce issues faced in both countries. 

Different countries, similar issues

When I was asked to write this blog my first thought was to compare and contrast the issues being faced in the healthcare sectors in the UK and Canada. Since October 2020 I have been CIO for the twelve hospitals in Northwestern Ontario in Canada, but prior to this I spent 7 years in various digital transformation and IT roles in the NHS in Yorkshire. This gives me an interesting perspective of how the two systems and approaches differ, but as I thought about it more I realised that today the major issues facing the NHS in the UK and the healthcare system in Canada are surprisingly similar.

Both England and Canada are facing critical shortages in the workforce. The lack of clinicians is impacting waiting lists and increasing pressure and workloads for frontline staff. This is leading to low morale and burnout. This creates a vicious circle with resignations further increasing workloads for those remaining. This situation is driving increased interest in technology solutions which can deliver rapid improvements to take the pressure off clinicians.

Reducing pressures on stretched workforces

Despite the introduction of electronic health records (EHR) over the past decade, the workforce shortages mean that the documentation burden on clinicians remains high. Increasingly ambient Artificial Intelligence scribes are being looked at by many organisations in the sector as a way to rapidly reduce the record keeping pressures faced by staff.

AI scribes record the face-to-face interactions between clinicians and their patients and generate written notes in real time based on these conversations. Iterative machine learning is applied to improve the accuracy of the transcriptions over time. By automatically creating text records, the technology has great potential to reduce the daily growing documentation burden on clinicians.

Harnessing the potential of AI

The potential benefits of AI scribes are obvious. Electronic records created automatically from conversations have the potential to greatly improve efficiency. Patient–physician encounters can be enhanced as the clinicians are able to focus on the individual and the verbal and non-verbal information, rather than concentrating on typing or writing what is being said. The actual conversation is captured, not just the clinician’s instant interpretation. The burden of catching up on documentation at the end of shifts or on days off is minimised. Indeed, some pilot projects are indicating that AI scribes could save up to half a day a week per clinician – improving efficiency and reducing workloads.

IT professionals in the healthcare sector can see the possibilities but I’m noticing that they are often reticent at jumping in feet first to widespread implementations. Clinician innovators are setting the pace as they can see the benefits which will accrue directly to them, however there are potential issues which are important to clarify and iron out before letting rip with the solutions across organisations and the sector as a whole.

Technology cannot replace clinical expertise

AI scribes are not a replacement for clinicians and their responsibility that accurate records are kept. The AI is transcribing from human speech and may struggle if the patient’s condition means the volume level is low, their speech is slurred or there is a high level of surrounding noise. Have the tools been ‘trained’ fully across different populations of people - nationalities, accents and cultural contexts? How is the interpretation being done, is it consistent and trustworthy, is there bias in the learning and how does it learn ongoing? An error in a condition, symptom or medication could have very serious consequences. Clinicians will need to review the content created so that inaccuracies can be corrected, so record taking will still take up a portion of their valuable time. It is important we don’t just spend money fixing one problem which then creates another.

IT professionals in the sector can also see that the right infrastructure needs to be in place as we move from pilot implementations to wider take-up. There needs to be safe and secure storage for an appropriate length of time. This is an additional type of data which could be susceptible to catastrophic cyberattack. Where will the data be stored? If this is in the cloud then the solution providers may hold the data in another country. This territory might have different data protection or retention rules, for example, which is unlikely to be acceptable for such sensitive information.

Iron out the potential issues first to secure lasting benefits

Based on my experience in digital transformation projects in healthcare in both the UK and Canada, I can see the great potential advantages that AI scribes offer and how numerous benefits could be quickly delivered to frontline staff and patients. However, I’m hopeful that we don’t just jump into AI solutions without the right certification of tools, understanding of legislation including how the patient fits into the use of these tools, clearly defined governance and scope of use rules, and robust infrastructures supported by appropriate budgets. A wider review of the ethics of using these AI solutions and whether patients have the right to access the recordings is needed. Will new protections and legislation be required?

AI scribes offer tantalising advantages but we need to look at the wider picture before diving in head first. A short amount of time ironing out the potential issues can still deliver the expected benefits relatively quickly and save time and money in the long run.