In the latest report from our HTN Health Tech Trends Series sponsored by InterSystems, we asked NHS professionals about artificial intelligence.
Since March 2019, we surveyed health care professionals to understand more about artificial intelligence to support health care delivery. In this article we highlight some of the responses.
When asked ‘do you see potential in using artificial intelligence?’ To no surprise, an overwhelming majority answered ‘yes’, with only 25% of respondents answering ‘no’. One theme highlighted from the results was the need for AI to assist or takeover administrational tasks and manage data – something which we have found to be a recurring issue time when interviewing health care professionals.
One respondent commented “There is huge potential with AI against a backdrop of 100,000 vacancies in the NHS. It needs to play a key role in removing repetitive tasks and providing quicker solutions to data analysis.”
It was also suggested that AI could have the potential to perform more complex tasks such as “prescribing chemotherapy” or “act on abnormal findings”. A CCIO commented “Demand for imaging is increasing rapidly, AI is able to prioritise certain images and also look at DNA patterns, all elements to support our staff and be more efficient”.
It was clear respondents highlighted the need for AI to “free up time for clinicians”, “automate routine tasks”, “do a lot of the routine correlation of data” and “save time and effort to use health care staff where they are most beneficial: doing what they need to do for patients, with patients”.
Great potential
Undoubtedly, AI has a great potential to fulfil in society. In an era of digitality, AI is changing the way we communicate and connect with technology. There are those amongst us who firmly believe that current research into human-computer interfaces can take us into a post-human era of total amalgamation between the biological and the technological.
As identified in previous interviews with clinicians, with many of whom strongly asserting that our current usage of AI is but the tip of the iceberg, there appears to be a collective understanding that AI in health care is the path to a brighter future.
There is of course a lack of understanding by some about what AI actually is, and understandably so. From a philosophical standpoint, ‘artificial intelligence’ implies some kind of ‘man-made self-aware entity’ where in reality, we are talking about complex algorithms within a system. One respondent summarised this point, saying “it is about understanding what is meant by ‘artificial’ intelligence”.
A business analyst working within a health care setting commented on the huge potential of AI from our survey: “There is a mass of potential in using AI, whether this is to be able to better plan and predict services, the effectiveness of treatments or when someone will benefit from an intervention; can we identify if someone is becoming frailer sooner and put something in place to prevent deterioration. Linking weather forecasts/pollution forecasts/pollen warnings to those that are most at risk from this. We are only just scratching the surface”.
A consultant in a large acute trust said “we must also understand where we are, and we must keep our feet on the ground, over hype is good for no-one. There’s AI technology out there with screening detection rates around 30% – the gap between reality and hype needs to be understood.”
Complex tasking with risks
Aside from dealing with administrative tasks, respondents from the survey commented that AI could be used for more complex procedures.
A lead pharmacist technician pointed out that AI “has some potential in being used to identify at-risk patients” with another respondent stating that “AI has provided a vastly improved accuracy of care to patients” and “AI augments clinical practice and supports diagnosis”.
One respondent highlighted the potential in supporting clincians digest information from within the medical record “AI could offer better ways of presenting data, such as using Natural Language Processing to identify clinical terms in narrative documents (progress notes, risk assessments, care plans), then listing these alongside the narrative, so that clinicians can understand the medical summary. AI could manage individual clinical records better – prompting users when a new assessment is needed, or a consent is expiring, even closing the record with automated measures if a death date is recorded.”
Away from these positives, it should be noted that the implementation and development of AI has already come at a price.
Touched upon by one respondent “If used correctly, it can help ease clinicians stress” with the operative phrase here being ‘if used correctly’.
Unfortunately, we have already found that AI has accidently created a racial bias in a part of the US health care system – a study into a widely used decision making software has found significant bias affecting millions of people.
The study published by Science found that the algorithm was less likely to refer black people than white people who were equally unwell.
The hospitals use the algorithm to guide care decisions, but the authors estimated that this racial bias reduces the number of black patients identified for extra care by more than half.
In this case the bias occurs because the algorithm uses health costs as a proxy for health needs. Less money is spent on black patients who have the same level of need, and the algorithm thus falsely concludes that black patients are healthier than equally sick white patients. Reformulating the algorithm so that it no longer uses costs as a proxy for needs eliminates the racial bias in predicting who needs extra care.
The researchers found that the algorithm assigned risk scores to patients on the basis of health-care costs accrued in one year. They say that this assumption might have seemed reasonable because higher health-care costs are generally associated with greater health needs.
It would appear then, from our survey and beyond, that there has to be a cautious approach to implementing AI in clinical settings despite the obvious positive effects and attributes it has already shown.
Further information
For further analysis and comments on artificial intelligence, HTN asked industry experts for their views, you can read the article here.
NHSX, the body responsible for delivering the Health Secretary’s Tech Vision has recently published a new report on the state of AI in healthcare. The report highlights where the organisation sees practical examples for AI and what the organisation is doing to build ethics and transparency into the use of AI.