Keynote address in Las Vegas mentions Tallaght University Hospital

TALLAGHT University Hospital was brought to the world stage last week, when it was mentioned in a keynote address in Las Vegas that was viewed by thousands of people across the globe.

The address was delivered by Amazon’s chief technology officer and vice president Dr Werner Vogels, who mentioned Tallaght University Hospital when discussing machine learning in radiology.

Dr Vogels told attendees at the AWS re:Invent 2023 conference last Thursday about how he had visited numerous hospitals to ascertain how machine learning may help with radiology.

One of the hospitals he visited was Tallaght University Hospital (TUH), where he met with stroke specialists to see if there was a problem that perhaps machine learning could provide a solution to.

“In this hospital in Dublin,” said Dr Vogels as images of TUH were projected on the screen behind him, “I spoke to one of the stroke specialists.

“He hammered on to me, saying every second when a person has a stroke counts. A stroke patient loses 1.9 million neurons a minute if they’re not treated – so quick treatment is crucial.

“That sounds like a reasonable, simple enough problem that I can attack.”

Dr Vogels met with Professor Catherine Wall and Dr Dan Ryan when he visited TUH last September, and his meeting with them helped to inform his project.

The project looked at data such as CAT scans for haemorrhages being uploaded to the AWS Health Imaging store.

The image is then assessed and is sent to the neurologist who is notified by text, and the radiologist will also be notified for prioritisation of the examination of the image.

“When I visited that hospital in Dublin, the neurologist said he’d rather get woken up at 3am for a false positive than not be woken up at all, because every minute counts,” said Dr Vogel.

“So he doesn’t want to wait until the radiologist gets through his re-prioritised worklist, he wants to get an SMS at night.

“At the moment my mobile detects that there is a brain haemorrhage, that’s evolvable architecture.

“Now that SMS can be sent to the neurologist and he can immediately take a look at the image, [and see] whether or not he should immediately jump in his car and drive to the hospital.

“It is that simple, it is not that hard to build these models. Anybody that’s been talking to you about MI, AI and ML and building these models and how hard it is – it is not that hard.”

Dr Vogels shared the rationale behind the project at the conference and added that he had built the model for it across five Friday afternoons.