By Iain Bray, healthcare architect, InterSystems.
The technologies of artificial intelligence (AI) are making significant strides in healthcare today, with an increase in the practical application of machine learning and decision trees. Globally, estimates indicate that the market for AI health technologies will expand at a compound annual growth rate of 38.5 percent from 2022 to 2030, by which time it will be worth US$208.2 billion.
In the UK, momentum is gathering too, with the Government having made significant investments in pushing the AI in healthcare agenda. The £140 million NHS AI Health and Care Award, for example, has funded a broad spectrum of AI health technologies at various stages of their development, with the end goal of accelerating innovation and bringing these technologies into routine use.
We are seeing pockets of functionality already being introduced also – either as trials or in real clinical settings. Examples include applications for image analysis for consistent pain evaluation, ADHD, skin condition, and Alzheimer diagnosis, as well as triaging waiting lists, clinic attendance, and readmission.
Quiet innovation from AI is becoming more prevalent also. InterSystems, for example, is trialling a Smart Assistant that improves user productivity by reducing keystrokes and enhances data quality.
Building on these early examples, work is progressing today across many settings in the UK to ensure that AI can be effectively evaluated, safely deployed, and scaled within the NHS. Much of this research is focused on helping clinicians to do their jobs more effectively, with the most mature technology being used in the diagnostics space.
There are multiple drivers for this growing usage. On the one hand, there is a sense that innovation is critically important to healthcare organisations, and in line with that a real desire to achieve it, as providers strive to deliver the best possible patient care in amongst ongoing resource shortages.
Underlining the significance with which innovation is regarded across the sector, a 2022 survey commissioned by InterSystems, found that almost three-quarters (71 percent) of healthcare leaders believe it is vital to the survival of their organisations. AI plays into this conversation as the enabler of innovation – a catalyst for change to help ensure optimum outcomes for patients.
Using AI in the right way
Clinicians need AI to help them do their job more efficiently, focusing on the types of technologies that will make their jobs easier and improve overall productivity. They want AI to help deliver actionable intelligence and recommendations that make workflows better, speeds processes up, and makes them more efficient. They increasingly appreciate, and want to leverage the fact that AI can also be used in training, with a focus on enhancing learner analytics.
However, a note of caution should be sounded here. As healthcare organisations seek to use AI to streamline efficiency and deliver better outcomes for patients, it is important they do not approach it from the wrong angle. AI should not remove decisions from the hands of clinicians – it should instead support them by tackling the laborious task of data-crunching. Only a human should make a clinical decision – but humans in healthcare often have limited time and struggle to analyse accurately an expanding mass of medical data.
AI should instead be complementary, taking care of the data analysis drudgery to provide real-time insight that improves the decision-making of highly-trained clinicians. Rather than dictating to them how they should work, it has to be about enabling clinicians to use their own skillsets so they take the right course of action for the patient every time.
These insight-driven decisions will be free of the distortions arising from excessive deference to hierarchy, over-reliance on custom and practice, or complacent groupthink, all of which have their biases. Having conducted the data-crunching element, AI allows human experts to spot what is missing from the available information and to make a decision that will result in a better outcome.
If it is to be successful, AI also needs to be easy to understand and to adopt. Familiarity is key. Clinicians need to be more comfortable with using AI, meaning it needs to be a ‘glass box’ not a ‘black box’.
Navigating change – what needs to happen
Introducing AI in this way is critically important because if they fail to do so, healthcare providers run the risk of alienating their teams of skilled clinicians and administrators.
Organisations must think holistically about the implementation of new technology. In particular, they need to understand the difference between technical and adaptive change, realising the latter can be more far-reaching.
Switching to a new drug or MRI machine is a technical change. It is something most healthcare professionals will be familiar with, and will have a well-trodden change pathway with little anxiety about the change going forward. Adaptive changes, in contrast, of which the implementation of AI is one significant example, are less clear-cut, more difficult to identify, and easier to disagree with. The process of digital transformation is an adaptive change, not a technical one.
It takes more than investing in the latest AI tools and technologies themselves to bring about true change – it requires a more holistic investment in the people and culture underpinning it, together with strong leadership. Providers need to get the support of the senior team who can help to drive the implementation forward. It is important to get operations staff on board also in order to execute that implementation. If that’s not acknowledged, understood, and addressed, any change will be difficult to implement.
Beyond this focus on cultural change, there also needs to be a broad investment in training and skills development. AI evangelists within the organisation need to start to support and augment the clinical and non-clinical workforce and bring skills up to scratch. AI learning will need to exist online and be accessible, so that clinicians can access it to continue their professional development.
There is much work to do in skilling up the workforce ready for accelerating the roll-out of AI. In the InterSystems survey, 38 percent of UK and Ireland healthcare leaders pinpointed skills gaps, which were cited by 38 percent of the sample as one of the biggest barriers to innovation within their organisation, making it second only to budget constraints (43 percent) in level of importance. These gaps often relate specifically to data, which many see as the fuel driving AI and advanced analytics.
It is also important to consider the user experience, of course, and that means patients. How do they want to be treated – and are they happy with AI technologies being applied to their data? The data stewardship question is fundamental here. Patients need to know who is managing their data, as they will almost certainly ask themselves – do we trust the institutions that are bringing the product to market and do we trust our clinicians – and they will need to be able to answer all that in the affirmative.
Looking ahead
AI is already here and positively impacting healthcare in UK and Ireland today. It is still primarily being applied in isolated instances, however, and much work still needs to be done in managing cultural change; building data skills; scaling up capability and finding use cases. The benefits of AI in healthcare are increasingly well understood but as more of the technology is introduced, healthcare providers will need help in streamlining implementations and optimising the benefits.
As AI usage ramps up, they will need to seek out opportunities to partner with organisations that have the expertise to help them identify the best and most appropriate applications. It is important ultimately that healthcare providers build an AI product or application that solves a problem people already want to solve, rather than building a product and convincing people it is a problem that needs addressing. That’s where partnering with an organisation with genuine AI capability and expertise can really help.
