So, despite forgetting, and being reminded a week before about this, I found myself on a ridiculously early (and expensive) train to London. I have an interest in surgery stemming from two weeks in a neurosurgery department. I also know AI will be becoming a huge part of my working life as it slowly takes over our world, so I was intrigued to find out more about it for the experts.
The day began with a talk by Dr Mori who was part of a team in Japan that had pioneered an AI that could identify polyps with high accuracy. It could also identify whether it was malignant or benign via a microscope at the end of the endoscope and sigmoidoscope.
The technology was amazing. It could identify polyps that a room full of experienced surgeons missed upon the first viewing. It could revolutionize the treatment of colorectal cancer by identifying abnormalities quickly. The technology compared thousands of scans that had been shown to the programme and the image from the patient. There is also a small microscope on the end of the scope and this can be used to further define wherever the abnormality was cancerous or benign, which was also determined by comparing the patient’s scans with thousands of preloaded scans.
We then had the Hugh Dudley memorial lecture which was one of the best lectures I have ever sat through. The topic was about death by PowerPoints. Something I was very familiar with. The speaker was Mr Terry Irwin and it was brilliant. I learnt some valuable skills about giving presentations that I hope to take into my medical degree and further teaching beyond my degree (I sound like I’m writing my personal statement here). A few of the later presenters were squirming in their seats as he pointed out everything that was wrong with PowerPoints and thus pointing out everything that was wrong with their PowerPoints.
After a quick tea break, we had the debate. The motion was “This house believes that AI causes a threat to patient safety”. I was against this motion as from the talk by Dr Mori in the morning, I couldn’t see why AI would harm patients, surely it would help improve patient care. The debate ran over dramatically but it was thought-provoking to hear the views of the speakers. A question was asked that I had wanted to ask when I arrived about the responsibility for any errors caused by the machine. Who would be to blame? This was quickly decided that it would be the surgeon as ultimately, they would be the ones in charge of the machine.
The closing remarks resulted in the split of 30% for 70% against swinging dramatically in favour of the motion. This was due to a comment from Professor John Fox about how if you believe medicine is more than facts and numbers, you should fear for patient safety with the introduction of AI. This is true as AI work on the principle of numbers and pre-calculated data. The swing ended up at 52% for and 48% against, reminds me of Brexit….
We then had the preliminary sessions of which speakers gave an 8-minute presentation of their own research. I was particularly interested in the first talk about neurostimulation and the performance of surgeons. Coming from a neuroscientific background, the idea that you can improve the efficiency of junior surgeons by stimulating the brain intrigued me. They found that the pre-frontal cortex of senior surgeons was significantly higher than that of junior surgeons. This is probably due to training, so they are focusing less on their fine motor skills and more on the situation at hand (this isn’t scientific knowledge, it’s just me rambling). They found that by using tDCS before surgery to stimulate the junior surgeon’s prefrontal cortex, operations were more successful. However, I am sceptical as they did not quote exactly how they proved the operations were more successful and when the question was posed, the guy seemed a bit flustered and could not exactly say how they knew the surgeons would be more successful. Perhaps something to further test on. (Look at me pretending like I Know what I am talking about! :D)
After lunch, we had a talk about using AR (Augmented Reality) in surgery. In particular, keyhole surgery in Liver operations. Liver operations are likely to be done via the traditional open surgery method due to the intricate nature of the liver and the vital vessels and tracts that lie within the liver. Professor Brian Davidson talked about how they were developing a system where the patient’s vessels could be projected onto their own liver, so keyhole surgery would be less risky. it was fascinating to watch the research develop to the model they had today. It had involved some pig livers, extensive programming and even some back to basics anatomy (as Pig Liver Anatomy seems to be absent from the medical school curriculum). If memory serves me correctly, they were ready to start to go into clinical trials, and I am excited to see where this technology leads.
We then had a second plenary session and finally arrived at the Simpson Smith Memorial Lecture given by Mr Manoj Ramachandran “Artificial intelligence in medical imaging”. This was a brilliant lecture to sit and listen to and inspiring to any future medics. During the lecture, a book was passed round in which we signed our names. This had been a tradition in every Simpson-smith lecture and you could feel the history of the book. I wish I had a chance to look through the book as apparently there were some famous names in there, to which I wonder if Henry Marsh is one.
The lecture was about using AI to increase the efficiency of hospitals and showed Mr Ramachandran’s journey of business enterprise to where he is today. The significant thing that stayed with me was his AI company viz.ai.
He has written 18 books on surgical education because ” when he was studying he knew he could write the book better”. So, it likely that most medical students will have read/own one of his books. The start-up that caught my attention was their Proactive Stroke Pathway (Link goes to the website where there is an animation).
What the pathway does is take CT scans from suspected stroke patients and sends them to the cloud to an AI which analyses the scans for signs of a stroke and if one is detected, it sends a message with the CT scans to the nearest stroke specialist. This reduces treatment time rapidly reducing the impact of the stroke on the patient’s life. It is the only FDA approved AI technology like this and is implemented in hospitals in the USA.
I was left speechless by this presentation. As someone with an interest in neurological pathology, I was keen to find out what this would mean for the NHS. However, I then began to think broadly about the implications of AI for teaching.
There is a risk we will begin to lose core skills. Why would we bother to point out easy to miss polyps to medical students if the AI is just going to do it anyway. Why bother to learn the detail of the liver if AI is just going to show us anyway. I understand that we need these skills, but will it impact teaching to undergraduates? In fact, this is becoming such a significant development that there is an EU wide meeting occurring at this moment to discuss the impact of the undergraduate teaching curriculum that AI will bring and how the curriculum needs to be adapted to incorporate AI.
The end of the day brought a lot to think about on the train home. I’m excited about the developing AI technology and what it will bring to patient healthcare, but I can’t help worry if it will have a negative impact on our teaching as future doctors. Will be get to the stage where we become mere machine operators?