A personal perspective of the AIMed Imaging, Clinician Series event by Dr. Elizabeth Morrow, Research Support Northern Ireland, UK

 

 The two-day online AIMed Imaging conference in June opened with a spectacular promotional video for next January’s AI Global Summit in California. This event will bring together 80 speakers from across the world, to change the world. The visuals and headline messages showed the cutting-edge of intelligent healthcare. It was very exciting and something I want to be part of in a small way, that’s why I write my reviews.

What struck me about this vision for where AI can go is that it isn’t just going to happen by itself. Although lots of people think AI is going to hit healthcare systems like a tsunami, I realised it is not about ‘riding the wave’ and seeing where it will take you. These people here at today’s meeting are the wave. They are the driving force for our future intelligent health systems. They are propelled by the experiences of people like Leanne West from iCAN (International Children’s Advisory Network), and patients like Nadia and Ed, who live with chronic life-limiting conditions and the hope of better solutions.

AI in healthcare will not just happen. It will take a million collaborative, informed and sensible decisions every day, to make an intelligent future real, and probably a lot of poor decisions, upsets, and failures along the way too. That’s why AIMed is so important for raising awareness, discussion, and networking across the health fields. These debates are also a public interest issue, which is another reason why I write about them.

Under the leadership of Dr Anthony Chang, the event is made possible by the support of sponsors (notably IBM) and the collaboration of academic and healthcare provider organisations. It is plain to see that for Dr Chang, AI is a passion, a commitment, and a personal interest that he has created a nexus for with AIMed.

Yes, the prospect of intelligent imaging is exciting and cutting edge, especially the idea of more precise and personalised health information and treatment and care solutions tailored to each person. But as the many excellent presenters went on to explain, there are hard problems, challenges, laggards, anxieties, and resistance in some fields that will have to be addressed in a sensitive, strategic, and evidence-based way.

Alok Gupta, guest speaker, from IBM, described the challenges of AI development, deployment, and adaption at scale. The most apparent, and perhaps greatest of these challenges, is the curation of impeccable interoperable health data systems. In plain English this means patients and practitioners need to be onboard with collecting accurate and useful data in ways that do not infringe on the patient/practitioner relationship.

While data challenges were mostly discussed in relation to the US Electronic Health Record (at least in the Health Executive stream) and what is needed at a national level, poor data will stop intelligent imaging being any use to anyone, in our countries and globally. This makes data quality a global issue. It is also a vital important professional issue because of the emotional toil for practitioners who have become ‘the most expensive data entry clerks on the planet’.

Data quality is a challenge that advances in ambient data collection and wearables could help to address. The big if is, if clinicians and patients trust what is in the ‘black box’ and how the data from these devices will be used. For me personally, I do not want an Apple Watch or a Fitbit or any other technology strapped to me, even if I was paid to wear it. However, my children might feel differently when they are adults. I might also feel differently if I were ill and could see the benefits of collecting my own data, for me personally, or to help other people in similar situations to me.

What I don’t want to happen, in health, is for people to be coerced into collecting data in the way the UK supermarkets and businesses are currently doing with their customer loyalty programmes. The deal/trick is to offer discounted prices in exchange for your customer data. My fear is that for people opting out of health data agreements, it could be a much more expensive or excluding, both in real terms and because of skews in data that exacerbate health inequalities for groups that are less likely to trust and sign up. But this is not an issue that is specific to AI Imaging, and will undoubtedly be a theme of the 2022 summit.

The presentation of specific ‘use cases’ is where AI’s capabilities for out-performing humans in perception tasks is displayed. Dipping between conference streams for cardiology, dermatology, ophthalmology, pathology, and radiology, I learnt that subspecialties are maturing in their acquisition of image data and negotiating how to use AI as a resource far beyond the tasks of clinical image acquisition and interpretation. With this convincing evidence of what AI can show clinicians and patients, the role of AI can be envisioned within the whole system of workflow, clinical encounters, patient experience, professional development, and so on.

My experience of quality improvement (QI) in UK health services tells me that certainly more could be done to hook AI into service provider’s existing QI work and Lean thinking programmes. As well as opening-up more and more effective feedback loops for service quality monitoring and improvement. AI could be the essential platform to meaningfully engage clinicians and patients in a digital web of interconnected health and wellbeing practices. Making more use of cloud computing for data sharing and edge computing could accelerate and streamline digital processes and gather data to show how whole organisations and territories are delivering health impact.

We have heard a lot recently in the UK and Europe about regulation of AI but for me a pressing issue is the professional education and development that sits behind all of this.

Digital professionalism, AI-assisted training, and digital ethics are particularly important and require more strategic engagement with education commissioners and providers to modernise curricula and competency frameworks. There is a global need to upskill the existing healthcare workforce about AI and to guard against further demoralising staff and dehumanising patients with data. Once again, AIMed will be an essential conduit for international collaboration and awareness raising about intelligence based interprofessionalism.

Amongst the excitement about the expansion and connection of specific use cases, I heard a measure of caution about not overdoing AI. For example, being aware of the inefficiencies of over-screening and over-diagnosis of non-fatal conditions. It shows that clinicians and healthcare provider organisations have an important role to play in deciding the most clinically effective and desirable applications of AI. There is clearly a role here for patients too, in maximising the ‘value added’ to their health and care by contributing their perspectives about the importance and impact of AI on their wellbeing.