Summary of my time at the UKIO congress 2022

The UKIO congress 2022 was held in Liverpool at the ACC arena, and contained three packed full days of lectures and exhibitions.

Overall there was a heavy focus on AI, however most talks were from industry talkers, attempting to sell a product. There were not as many discussions on how this is being implemented already, or the impact it is having. There were a number of really interesting papers discussed, and more information will hopefully be available in the next few years, which makes it clear that AI is still very much in its infancy.

Some of the challenges discussed with implementing an AI program within healthcare included the difficulties with ‘overfitting’, where the use of a dataset from a smaller community (such as a single hospital or platform) causes the AI to become too attuned to the data on which it was trained and therefore it loses its applicability in other areas. For use in healthcare the software needs to be generalisable, and adapt to different scenarios. Standardisation and good data across systems can really help with this. People are not identical, and this causes variation in how we report and how we record data, which makes it much more difficult to train an AI. Making sure our data is accurate and useful is also really important. The adage ‘Garbage in=garbage out’ was spoken several times.

The discussion around how AI may influence healthcare education was also very interesting. There is a school of thought that AI could be used to tailor education to the individual, increasing learning in areas individual students find most difficult, and vice versa. There are also early discussions being had regarding using AI to distinguish when a student has enough knowledge in their field to pass their course, this could be used in conjunction with exams and may allow universities of the future to pass students early, freeing up lecturers time. All of this is very interesting, and would totally revolutionise how we go about higher education in this country, however it is a long way off.

One of my favourite lectures of the congress was “Intended and unintended consequences of Al in paediatric radiology” by Susan Shelmerdine. She explained AI and machine learning in a simple and clear way, and gave interesting examples of current uses, including a video on how the AI software DALL.E can produce a unique digital image based on typed instructions. She spoke about how AI could help in paediatric radiology. This included:

1. More efficient workflow
2. Improved image acquisition
3. Interpretation/Accuracy
4. Automated measurements
5. Prognostication

Finally, I attended a lecture on the future of pre-registration education. This was pertinent to my current position. The first speaker discussed the upcoming revision of the HCPC standards of proficiency and how they may affect our teaching. The discussion then turned to the development of common clinical assessment tools for education. Interestingly there was a suggestion made that we need to reevaluate the balance of conventional x-ray and cross-sectional modality teaching, with focus shifting to more of an even split. This makes sense due to the focus of clinical practice sites shifting to those areas, meaning radiographers will be expected to know more about them. There were concerns raised about the quality of images being taken in this DR age, and more focus needs to be on image evaluation. Moving away from the imaging side of radiography, there is also a push for radiographers to become more rounded professionals upon leaving higher education, with governance, finance, competency documentation and performance being made a priority.

Clinical assessment training is needed nationwide, there is a severe lack of assessors, usually due to lack of, or unavailable, staff. It’s a well known problem that staff are reluctant to give negative feedback, and students often respond negatively if they do. With more training and structure we can create better working relationships. The Northwest team have developed an asynchronous online training module, and are looking at creating eLFH modules that may be available to everyone https://portal.e-lfh.org.uk/Component/Details/733573. This would be incredibly useful for the clinical team and I will be passing this on.

Overall I had a thoroughly enjoyable time at the UKIO congress, and hope to go again. It was admittedly exhausting, but I learnt so much valuable information and have come away with ideas for my current practice, for future teaching, and potential ideas for future research.