Last week I attended the British Science Festival run by the British Science Association, this year at Hull University. It was a week of wonder for all those interested in the world around them. From the secrets of the canine mind to using waste aluminium to capture carbon, there was something for everyone at this years festival.
I was fortunate enough to be attending the festival as a press intern, which meant reporting on all things festival. The experience itself was inspiring but better than that, it was an opportunity to learn.
Out of the many scientific revelations I discovered over the week one of the most interesting to me was about the future of medical technology. I was given the opportunity to write a blog post on my findings so I thought I’d share that with you this week. If you’re interested in hearing about other festival events and activities don’t forget to check out the British Science Associations website!
So here goes…
British Science Festival – The robot will see you now: the future of medical technology
by Alicia Shephard, British Science Festival
From iPads to space travel, robots to facetime, many of the predictions of A 2001 Space Odyssey have materialised in society over the past 50 years. However, one prediction you might have missed is metabolic monitoring during intensive care. This is something which is soon to become a reality thanks to the pioneering work of British Science Festival Award Lecture, Gemma Bale.
Gemma’s research is helping advance diagnosis techniques for the baby brain injury known as Hypoxic Ischemic Encephalopathy (HIE), a type of brain damage which occurs when a baby’s brain doesn’t receive enough oxygenated blood.
Due to the nature of the disease, it requires immediate medical attention including an MRI scan. However, the MRI can’t be performed until the infant is a week old, leaving a large window of time for the babies’ condition to worsen.
All hope is not lost though thanks to CYRIL, a machine which is capable of using colour to identify oxygen and metabolic levels within the brain. This is enabled by the translucent property of our own body which allows red light to pass through and be absorbed by oxygenated blood cells and the enzyme responsible for using oxygen in our metabolism. This light then ‘bounces back’ for detection and is converted into measurable metabolic and oxygen levels in the brain.
The levels of these components is highly coupled with the severity of brain injury. Therefore, this research could lead to identification of brain injury before an MRI scan would occur, ultimately resulting in more rapid, specialised treatment for the infant.
So, a 50-year old prediction is finally making its way into medical technology. But what do the next 50 years hold for us? Artificial intelligence is likely to hold a lot of the answers we’re looking for.
While it might sound like scary stuff, artificial intelligence is far from it. It simply means the ability of a computer to perform a task that usually requires human intelligence. If you really boil it down, something as simple and commonplace as a calculator could be considered artificial intelligence. Not so scary now is it?
A pathologist at work (Picture: JBSA, Staff Sgt. Jerilyn Quintanilla)
As Darren Treanor, a consultant pathologist explained at The robot will see you now event: “Artificial intelligence is used in all aspects of computer science and in recent years it has become much better at doing what we can do.”
This improvement is largely thanks to new techniques of teaching computers. Rather than using standard algorithms we now use deep conditioning methods which require vast amounts of data but are far more effective.
The use of artificial intelligence within pathology is a growing field of research. Between 2009 and 2016 the number of people waiting for a diagnosis doubled and there simply aren’t enough pathologists to do the job. Whilst it’s unlikely artificial intelligence could replace the experts altogether, they could rapidly decrease wait times and improve accuracy of diagnosis.
An experiment surrounding the precision of pathologist’s diagnosis found that, when given unlimited time to detect tumours from samples of tissue, the experts were able to do so with at least 95% precision. However, when asked to complete the same task with only a minute per sample, the accuracy dropped dramatically to as low as 50%. Pathologists experience pressure just like this on a daily basis. The advantage of using artificial intelligence is that it doesn’t experience these same pressures but is still capable of achieving the same results.
The overriding consensus is this: a computer can make a diagnosis much faster than a human, and has the advantage of being unaffected by health, fatigue and emotional influences. But any work it does should always be monitored by an actual person.
Now that all sounds very reassuring, but what about the dangers of artificial intelligence getting it wrong? Not to worry: pathologists, software engineers and artificial intelligence experts alike agree that this technology should not be used alone. In its current state it’s a method of increasing rates of diagnosis.
Who knows what predictions we’ll be making in 50 years’ time, but for now improving diagnosis through innovative technology and artificial intelligence appears to be the path to take.