AI explained: How machine learning could save our healthcare system
- Articles, Blog

AI explained: How machine learning could save our healthcare system


A system stretched to breaking
point. Perhaps the solution lies not with more doctors, but with machines
powered by artificial intelligence. We are very quickly going to reach a
tipping point where AI can transform health care.
This is a research project at the moment. But the
potential is enormous. For the NHS,
AI could mean better care and lower costs as machines take on some of
the jobs done by doctors their power comes from not just analysing images
in milliseconds but learning from the data they collect. Deepmind Is
leading the way in AI not just in the UK but in the world. It was set
up by three partners in 2010. Four read years later it was bought by
Google for £400 million. — four years later. This gives us a sneak
preview into a new type of health care that patients laugh. But it
also unnerves people worried about medicine slipping into corporate
control. One of the founders talked to us exclusively about the
company’s project with the NHS. What
I am really worried about is that the fear and the reactionary
paranoia is going to limit the access to what is clearly going to
be an incredibly valuable technology which will change people’s lives.
In the UK the total cost of blindness
is £28 billion every year. Equal to a fifth of the health Budget. Every
220 people go blind because of macular degeneration, a completely
treatable condition when caught in time. The Royal College of
ophthalmologists says patients with and acute version of the condition
should be seen within two weeks. At Deepmind the technology they are
pioneering is getting machines to teach itself. Last year, in a
ground-breaking moment, a computer out smarter the best human brain in
the ancient Chinese game of Go, the holy grail of artificial
intelligence. This game is special because there are more possible
moves than they are atoms in the universe so there is no way of
calculating every option on the board. Instead, the machine have to
mimic, learned from experience, a little bit like the human brain.
Every time it did this it got better and eventually it beat the Grand
Master. Now the same technology is being used to save people’s site. LA
knows all about the trauma that can bring. 15 years ago she started to
go blind in her left eye. — Elaine knows all about the trauma.
I remember it. I was walking along a
path in the woods. The shadows were getting darker and darker.
Everything was getting dimmer and dimmer. And I felt less than a
person. I felt… I felt alienated from everybody else.
Then just three years ago Elaine faced the prospect
of losing the site and have the eye, as well.
I waited maybe six to eight weeks for an appointment, worrying
all the time that something dreadful was going to happen to this I. Just
thinking about what the consequences would be if you Bartholdi while
blind you are vulnerable. And life for me, then, would not have been
worth living, it really wouldn’t. Thanks to AI it has turned out to be
a very different story for her. She was one of the first AI patients at
Moorfield. And instead of living in the shadows she spends her time in
the clouds. She is raising money for other patients facing blindness.
How was it?
Absolutely fantastic. This
is AI in action. It looks like a normal scanner but inside there is a
machine learning a tool that can analyse thousands of complex images
almost instantly. They show not just the back of your eye but the
cross-section of the retina at a higher magnification than an MRI
machine. This is part of your
central nervous system… Hours of
work for a person, just a few seconds for a machine.
A common cause for blindness is AMD. The
thing about it is that 200 people every single day, just in the UK,
get the blinding forms of it. The Royal College of ophthalmologists
suggest these patients should be seen and treated within two weeks of
the onset of their symptoms. The reality is that on the NHS that
target isn’t being met. AI would allow us to identify those patients
and get them treated not just within the two weeks, but potentially
within two days. Artificial
intelligence is a remarkable achievement. You can see that in a
place like this, but with it comes some very serious dilemmas. Will
people ever accept mistakes made by machines because there will be some?
And when those mistakes take place in who is responsible, the machines,
or the people behind them? There was
a problem over lack of clarity and liability issues when you get into
things like machine learning. If you take something like health care,
within a short space of time it may become negligent for a doctor not to
use and AI aid. But in that situation who is to say who is
liable? Is it the machine ‘s fault, or visit the doctor who hasn’t
really assessed that the outcome of the tool is valid? These are
difficult issues. It always has to become in my view, the doctor Who
remains liable. We’ve already seen it in an interesting experiment in
Sweden where some sort of machine learning bot was let loose on the
dark web. It was given 100 euros per week to go and spend on the dark
web. It went off and bought illegal drugs, firearms, and surveillance
equipment. The authorities found out about this thing and came in and
arrested it. Of course, you cannot arrest a machine, but we hear these
things, and ascribe liability to machines, when, of course, the
things being done really the responsibility of the people who set
spot free. AI is now being used in
specialities beyond ophthalmology despite the doubt. At this hospital
it is being trialled in both heart surgery and pioneering feat all
scans. What artificial intelligence
allows us to do is to capture the training and experience of thousands
of people very quickly. — and pioneering foetus scans. We captured
a lot more than any single person could capture in a lifetime.
For Rebecca Hooper and many other
mothers, these scans show details never seen before.
A lot of the systems we have are overwhelmed. We
are talking about millions of pictures being created at each
centre where the scanning is being carried out. It’ll help you to see
if there is anything wrong with the baby.
If AI can do all that is this the end for doctors?
You have to think about the systems as being
tools. Just like scanners and scalpels. They are assistants which
help humans to do better. As a
creator of the technology he is a believer, of course, but despite the
progress at Moorfields Deepmind has been brought into controversy after
a hospital was given access to the information of patients and doctors.
But it raised doubts about whether such a big project should be awarded
to a commercial organisation, despite it having nothing to do with
AI. It leads to some questions about privacy for Deepmind.
We can protect and store this data. In addition, we
subject ourselves to the highest level of oversight. We have a panel
of independent reviewers who we have invited proactively to scrutinise us
in the public interest. We’ve given them a Budget to check out what we
are doing, interview our staff, and hopefully they will publish a report
at the end of the year. You accept
that there is a concern that the information could be used
commercially? We’ve always invited
that the information could be used commercially? We’ve always invited
that’s gravity. But there are enormous benefits. — we’ve always
invited that scrutiny. Those
organisations stand to make a fortune. Economists now describe
data as the new oil. A 21st-century commodity which will be the driver
for wealth of companies. The question is will this be at the
expense of individuals? Do you think machines will help humanity, or will
it entrench inequality? We have to
think sensitively about any new technology that is introduced. If we
don’t introduce it wisely able entrench the existing order, put it
that way. If we care about equality we should definitely start to think
now, which I think is early, and that’s the right time, who benefits
from that technology and how you can ensure those benefits are as widely
available as possible. AI technology
can seem like it has been borrowed from a sci-fi script, but this type
of medicine is coming faster than you think. There are huge gains to
be made by patients and taxpayers alike, but there are risks, too,
because those gains depend on us giving up some of the secrets of our
medical history to corporations we may not be able to control.

About Bill McCormick

Read All Posts By Bill McCormick

4 thoughts on “AI explained: How machine learning could save our healthcare system

  1. The masonic eye of providence. The masonic G grand architect in the eye. Disgusting symbolism. Disgusting transhumanist filth

  2. Great description that shows how AI could improve patient outcomes avoid costly incidents and save lives.

Leave a Reply

Your email address will not be published. Required fields are marked *