17/12/2016 Click - Short Edition


17/12/2016

Similar Content

Browse content similar to 17/12/2016. Check below for episodes and series from the same categories and more!

Transcript


LineFromTo

Now, in a short while it'll be time for Newswatch.

:00:00.:00:00.

The Design Museum in London has moved into a new home,

:00:00.:00:29.

I have come to see Fear And Love, an exhibition of 11 designers'

:00:30.:00:40.

reactions to our increasingly complex world.

:00:41.:00:45.

The most animated star on show has to be an industrial robot arm,

:00:46.:00:49.

which its owner hopes will present a more friendly face to robotics,

:00:50.:00:55.

and even maybe help us empathise with mechanoids of the future.

:00:56.:01:00.

It senses where you are and comes bounding over to see you,

:01:01.:01:03.

but if it gets bored, it will turn its attention

:01:04.:01:06.

It is a bit like an excitable puppy, actually.

:01:07.:01:12.

Who knows, installations like this may help to allay our fears

:01:13.:01:15.

of being around giant machines like this.

:01:16.:01:19.

I have to say, it will still be a while before I trust this thing

:01:20.:01:23.

That said, computers are increasingly being used

:01:24.:01:27.

There is plenty of research into how artificial intelligence can help

:01:28.:01:33.

doctors to better look after patients.

:01:34.:01:37.

Jen has been taking a look at some of the latest developments.

:01:38.:01:42.

Around the world, hospitals are facing a backlog of patients,

:01:43.:01:45.

ageing populations and a shortage of specialist staff.

:01:46.:01:50.

Some hospitals are teaming up with artificial intelligence

:01:51.:01:52.

research teams to see if there are ways that high-tech

:01:53.:01:55.

solutions can supplement or even enhance healthcare in the face

:01:56.:01:58.

Its health minister says they will need more than 30,000

:01:59.:02:06.

new nurses before 2020, and completely rethink the way it

:02:07.:02:10.

So when the CEO of one of its largest private hospital

:02:11.:02:17.

networks approached IBM's Watson team, they came up with a pilot

:02:18.:02:20.

project to try to help nurses working with the most

:02:21.:02:23.

This is the intensive care unit at Mount Elizabeth Novena Hospital.

:02:24.:02:30.

It's where four beds are conducted to IBM's artificially

:02:31.:02:32.

Collecting all the vital signs from the patients in the beds,

:02:33.:02:38.

it gives the nurses a more complete picture of who needs the most care.

:02:39.:02:43.

In one of the first trials of its kind in the world,

:02:44.:02:46.

the AI is constantly monitoring output and making connections

:02:47.:02:48.

on a vast range of data, including a commonly used scale.

:02:49.:02:54.

Higher scores correspond to a higher incidence of death,

:02:55.:02:57.

and it is particularly important in the first 24

:02:58.:02:59.

This patient has four alarms, so if you don't see anything

:03:00.:03:07.

flashing here, it means it has been acknowledged already.

:03:08.:03:11.

One of the patients in the ward is at the high end of the alert,

:03:12.:03:15.

and nurses can quickly access the information in real-time

:03:16.:03:18.

and look at patterns in their vital signs to see if they are at greater

:03:19.:03:22.

Here in the UK, it's the help AI could provide in imaging

:03:23.:03:28.

between the NHS and Google's DeepMind.

:03:29.:03:33.

The UK's Royal College of Radiologists says 99%

:03:34.:03:36.

of hospitals are struggling to keep up with demand,

:03:37.:03:39.

and the UK has the third lowest numbers of specialists who can

:03:40.:03:42.

The large amount of data is overwhelming a health service

:03:43.:03:50.

If you can use algorithms or machine learning or artificial intelligence

:03:51.:03:58.

to set an alert for you, to trigger to say something has

:03:59.:04:01.

happened, you need to go and see this, this is urgent and you need

:04:02.:04:05.

to deal with it, in the next hour or so when you may have not

:04:06.:04:09.

I think it will improve quality of care and actually improve equity

:04:10.:04:13.

One of the first areas where the NHS is testing artificial intelligence

:04:14.:04:21.

is at Moorfields, one of the busiest eye hospitals in the world.

:04:22.:04:26.

DeepMind is applying the same machine learning technology

:04:27.:04:29.

behind its winning AlphaGo computer programme.

:04:30.:04:32.

It beat the world's best human player by computing tens

:04:33.:04:36.

of thousands of positions per second.

:04:37.:04:40.

We started DeepMind to develop general-purpose learning algorithms

:04:41.:04:43.

and use those tools and systems to make the world a better place.

:04:44.:04:46.

It was obvious to us a few years ago that there is a massive opportunity

:04:47.:04:50.

to deliver really meaningful and proved benefits to many patients

:04:51.:04:53.

and people across the world using our sort of techniques

:04:54.:04:56.

to try to improve the way we diagnose and treat patients

:04:57.:04:59.

The Moorfields Hospital research is using scans from this OCT,

:05:00.:05:08.

or optical coherence tomography machine,

:05:09.:05:10.

which creates a 3-dimensional retinal image.

:05:11.:05:14.

It is used to diagnose diseases like age-related macular

:05:15.:05:18.

degeneration, and diabetic retinopathy, two leading causes

:05:19.:05:20.

DeepMind is trying to develop a computer algorithm

:05:21.:05:26.

which will identify scans of concern.

:05:27.:05:29.

OCT scans were chosen because of the high rate

:05:30.:05:32.

of information included in them and the way

:05:33.:05:34.

they can be broken down into pixels showing areas

:05:35.:05:37.

I was particularly attracted to speaking to DeepMind

:05:38.:05:41.

because I thought their algorithms would have the best ability to deal

:05:42.:05:44.

with 3-D imaging of an extremely high resolution form,

:05:45.:05:47.

This is such a delicate area of the eye that any sort

:05:48.:05:55.

of disruption of the normal architecture has really amazingly

:05:56.:05:57.

So I believe health career could be at a pivotal moment in history

:05:58.:06:06.

where these advances in technology, such as artificial intelligence,

:06:07.:06:09.

will fundamentally change the way medicine is practised,

:06:10.:06:13.

If you think about it, the best humans in the world

:06:14.:06:20.

will have seen only a fraction of the number of cases that we can

:06:21.:06:24.

Imagine that we took all of the cases that

:06:25.:06:28.

many of the top ophthalmologists in the world have seen themselves,

:06:29.:06:31.

Now the algorithm can sample from all of the case studies

:06:32.:06:37.

that our various different humans have seen, and try to deliver a much

:06:38.:06:41.

higher standard, more consistently, when making a diagnosis.

:06:42.:06:46.

All these projects are still in the research or pilot stage,

:06:47.:06:49.

but it's fascinating to see how artificial intelligence

:06:50.:06:51.

could transform healthcare and perhaps lead to faster

:06:52.:06:53.

Meanwhile, back here at the Design Museum in London,

:06:54.:07:11.

some of the most beautiful 3D printing I think I've ever seen.

:07:12.:07:14.

These are one artist's suggestion about how we might revive

:07:15.:07:17.

the ancient culture of making death masks.

:07:18.:07:26.

I wouldn't mind one because it would make me look like I was in the film

:07:27.:07:30.

Next, we're going to ask - what would happen if you scaled that

:07:31.:07:35.

What if you were to let it loose on our homes,

:07:36.:07:39.

The buildings around us don't look the way they do by accident.

:07:40.:07:47.

The design, the shape and the structure are

:07:48.:07:52.

all the result of the mix between the desire of designers,

:07:53.:07:55.

what we need the buildings to do and the practical limitations

:07:56.:07:58.

of the materials and building techniques we've discovered.

:07:59.:08:01.

This is very much the age of concrete, steel and glass.

:08:02.:08:05.

But with new technology and techniques, what could the next

:08:06.:08:08.

The building industry is still in 19th century technology.

:08:09.:08:16.

It hasn't really evolved like other disciplines and if you look now

:08:17.:08:22.

at the speed at which cities are growing, so many people

:08:23.:08:25.

are moving to the cities, but our technology

:08:26.:08:28.

Industrial scale 3D printing has already been put to use to print

:08:29.:08:32.

full-scale buildings, like this housing project in China.

:08:33.:08:34.

But researchers are now turning to computers to not just create

:08:35.:08:37.

This is a prototype column that's been 3D printed here at the school

:08:38.:08:47.

of architecture at the University College London.

:08:48.:08:51.

We basically use a computer and use algorithms to generate

:08:52.:08:53.

They may look very alien and strange, but actually

:08:54.:09:00.

So these forms attempt to save material and become more

:09:01.:09:04.

efficient, but at the same time they produce a sort of aesthetic

:09:05.:09:07.

that is very appealing to us as architects and that really

:09:08.:09:11.

doesn't look what a normal building any more.

:09:12.:09:15.

Normal 3D printing creates objects by building up thousands

:09:16.:09:17.

of very thin layers, which you can imagine takes

:09:18.:09:20.

The idea here, though, is to save time by printing just

:09:21.:09:29.

what you need, which means rather than printing

:09:30.:09:32.

flat layers, instead built with shapes, like pyramids.

:09:33.:09:34.

The software they've created can take this a step further,

:09:35.:09:37.

by figuring out which bits are structurally

:09:38.:09:38.

essential and then getting rid of the rest.

:09:39.:09:43.

Before computers we had to build by hand, right?

:09:44.:09:45.

And now we can create algorithms that make these

:09:46.:09:47.

calculations for us, but that doesn't mean that we don't

:09:48.:09:50.

design, we just optimise the process more and we can create things

:09:51.:09:53.

that we couldn't ever think of before.

:09:54.:09:55.

3D printing will allow architecture to be much more detailed,

:09:56.:09:58.

much more fine, and also much more efficient.

:09:59.:10:04.

Like, if you can 3D print exactly the material that you need

:10:05.:10:07.

in a specific part of the building, it will make it perform

:10:08.:10:11.

Before these new techniques can be put to use, they first need to be

:10:12.:10:19.

Case in point, this MX3D Bridge project aims to 3D print a usable

:10:20.:10:26.

steel bridge right in the centre of Amsterdam.

:10:27.:10:30.

Created using similar generative algorithms,

:10:31.:10:32.

the project has been held up while the company proves

:10:33.:10:34.

to regulators that the design is structurally sound.

:10:35.:10:43.

The actual bridge now isn't slated to appear until next year.

:10:44.:10:46.

Techniques like these certainly promise to spice up our city's

:10:47.:10:49.

skylines, but it could still be a while before we see 3D printers

:10:50.:10:52.

That was it for the short edition of Click at the design Museum. You can

:10:53.:11:10.

catch up on our player. Next time, it is the Click Christmas party. Be

:11:11.:11:16.

prepared for anything. Plus a look at our best bits third 2016. Follow

:11:17.:11:23.

us any time on Twitter. Thank you for watching ours. See you soon.

:11:24.:11:26.

Download Subtitles

SRT

ASS