Every year,

700,000 people are killed by superbugs

To put that in perspective,

that's more than the combined populations
of the cities of Bristol and Bath.

What's more is that in about thirty years

that number is expected to rise
to 10 million people a year,

which would mean that superbugs
would have killed more people than cancer.

Now, many of us may have heard 
the term 'superbug'

thrown around at some point in our lives.

But lets take some time to think about
what it is we're really referring to,

how we got here,

and most importantly,

how do we get ourselves
out of this crisis?

So what is a superbug?

'Superbug' is a name given
to a particular group of bacteria,

otherwise known as a strain,

that cannot be treated
with most of the antibiotics

available or in use today.

They're a formidable enemy
when you consider the fact

that many people would not have lived
to see their sixtieth birthday

prior to the discovery of penicillin
and other classes of antibiotics.

Raise your hand if you've ever had
an antibiotic prescribed to you.

OK.

Well, you'll be familiar with the process
I'm about to describe next.

It probably all started
when you weren't feeling very well:

a persistent cough, a fever,

or perhaps that burning sensation
when you'd go for a wee.

(Laughter)

Having put up with it all weekend,

you finally mustered up that courage
to go out and speak to your doctor.

After reviewing your symptoms,

you're given a prescription
for antibiotics and told:

'Come back after the third dose
if your symptoms persist'.

Now, I'm an engineer,
so I love my flowcharts.

And they really help me
get my head around what's going on.

So I took the liberty of putting 
this little schematic together.

So step one: bacterial infection starts.

Step two: signs and symptoms develop.

Step three: person goes out 
to get medical attention.

Step four: antibiotic
prescription is given.

And step five: after three doses,

check whether symptoms are getting better.

If yes, carry on.

If no, order a diagnostic test 
to determine what is making you sick,

and then return to step three.

Now, I've got the utmost respect
for our friends in the medical profession,

but this system is fundamentally flawed,
for a couple of reasons.

Firstly, it encourages your doctor
to start treatment

before having all the information.

Secondly,

each time we go around that loop,

what we're actually doing
is killing off all the bacteria

that cannot defend themselves
against a given antibiotic,

leaving behind a monolithic group
of bacteria that can.

In effect, we're helping the enemy
sift out unfit soldiers.

Now, we're in this predicament
because, still today,

one of the most commonly used methods
for identifying bacteria

involves taking a sample of urine,
of mucus, of blood,

and growing it under varying conditions
to help us piece together the identity.

Think of it like a process of elimination
carried out in the lab.

Once the bacteria is identified,

we then use another process

to determine which antibiotic
most likely will work.

The problem is that that takes time,

two days or more,

and in some extreme cases,

people have actually died
before their doctors got the answers.

Nevertheless, this is our system;
treatment first, diagnostic second.

So how do we get ourselves out of this?

Actually, it would be a lot better

if we could first work out
what's making you sick,

and then selecting
the most appropriate antibiotic.

Not only would you get better days sooner,
but we'd also curb the rise of superbugs.

But to do that, 
we'd need a test that's fast.

I mean, like, really, really fast;

fast enough to meet the 20 minutes or less

that you have with your doctor
or your pharmacist.

But also, it'd also need to be affordable;

affordable enough
that developing countries

could make that transition.

And finally, it would need to be
easy enough to use

that it's as effortless
as checking your temperature.

And that is exactly what my team and I
have been working towards

over the last two years.

We've been developing a technique
based on receptor-mediated sensing.

We've quoted it GMS for short,
and it's a lot simpler than it sounds.

For many bacteria,
the first step of infecting you

involves sticking themselves
to the cells in your body

using a hook and loop system

that's quite similar
to what we've come to love in Velcro.

(Laughter)

The hooks in this case
would be specific proteins, or receptors,

on the surface of the bacteria,

and the loops would be specific molecules
on the surface of the cells in your body.

What we've done is created a low-cost,
light-emitting material

and coated it in loops;

let's call them probes.

When we mix our probes with a sample -
for example, urine -

they stick to bacteria like Velcro.

And by measuring the light 
coming off of them,

we can then calculate the number
of bacteria present in that sample.

The final process itself
will be quite simple.

Take a bit of urine, 
add it to a special cartridge.

Then place that cartridge
in our very own bug detector.

The cartridge would mix
the urine with our probes

before separating out the bacteria
and measuring the light coming off.

That final step will allow us to determine
whether bacteria is present,

which one, and how many,

all within 15 minutes.

From two days, to 15 minutes.

Now, as with most scientific developments,

the journey is fraught with challenges
and sometimes disappointments, frankly.

And I'll share a couple of them with you.

One of our first challenges was: 
how do you develop a probe

that effectively mimics
the cells in your body?

It took a team of researchers at the 
University of Bristol

months of experimenting

to get the recipe just right

to have the right type and balance
of loops to mimic the cell.

I mean, this is not like bashing out 
bad pancakes on a Sunday morning.

This is the kind of effort
you put into winning Masterchef.

(Chuckling)

The second challenge for us was actually
finding a detection method

that was low-cost, yet powerful enough
to detect the probes

once attached to the bacteria.

For that, we turned to a well-established
technique based on fluorescence.

Fluorescence is a phenomenon

where a material stimulated with light
absorbs some of it,

and re-emits a different colour.

As a detection technique,

we'd effectively take
a specific colour of light,

shine it at that material, 
and observe the colour coming back,

and that helps us tell what's present.

So, we've got probes, 
they stick to the right bacteria,

and we can detect them.

Job done, right?

Actually, not quite.

Most of the foods that we eat 
actually get broken down

into light-emitting materials
that end up in your wee.

And so, these are things like 
energy drinks, vitamin supplements,

pregnancy supplements,

and all of these together
could lead to false positives.

In other words, they make it really hard
for us to determine the difference

between friend or foe.

And so, differentiating our probes from
the remnants of your last energy drink

is absolutely important

for ensuring that we do not report
that bacteria is present

even when there aren't any.

This is where the research
took a really interesting turn,

because getting around this
meant getting familiar with our probes,

but also getting
really familiar with urine.

We carried out several experiments 
observing how our probes behaved

when stimulated with
different colours of light.

Particularly, we're interested
in how much green light is emitted

for every colour of light
that we stimulated them with.

This is the profile of our probes,

and it does not change.

So any deviation from that
would then tell us

that there's other stuff
mixed in with the sample,

and it would look a little bit like
when it's mixed with urine, for example.

By measuring that change, 
we can actively correct for interference

from anything else
that's present in the urine.

So pulling it all together, 
the probes help us find the bacteria

and fluorescence helps us find the probes.

We're still at the
early stages of our journey,

but we were able to take
our first prototype into a hospital lab

where we tested for E. coli 
in human urine.

GMS was able to detect the presence 
of urine without any false positives,

and it correctly reported the absence 
of E. coli in the urine in 63% of cases.

Finally,

by placing three measurements in parallel,

we were able to achieve an average time
per test of just four minutes.

This is certainly not
the end of the story for us.

We're now developing other forms of loops
to address other harmful bacteria,

and we aim to test and improve
the system a few more times

before it gets to your GP.

But once we do this,

combining our faster, accurate diagnostics

with better antibiotic usage
and continued public engagement,

we could finally tip this war
back in our favour

and overcome superbugs once and for all.

Thank you.

(Applause)