So we all know surgery.

Most of us will undergo surgery at least once.

They save a lot of lives.

They improve the quality of many other lives.

But the way we do surgeries today is very far from ideal.

Just to think of it, they're very, very expensive.

Training the staff, building the operation theater,

maintaining both, is a huge financial burden.

And that leads to the fact that most people on the planet who

need surgery don't have access to those facilities

and that personnel.

Just to give you some numbers, here in the US alone,

more than 50 million inpatient surgeries will be performed.

The surgeries will cost between $25 and $150 per minute

in the OR, not counting procedure-specific costs.

And between 1 and 3% of the patients

will die within 30 days of the procedure from complications.

What I'd like to do here is to propose to you

an alternative approach which could revolutionize

the entire field as it looks today.

So imagine that we could take all the facilities, all

the equipment, all the knowledge required

to perform a successful surgery, and encode it

in a single drop of saline.

So that drop can be put inside a syringe.

You can take it anywhere on earth.

You can give it to anyone in need

without even having to anesthetize them first.

Inside the patient, the drop inside the syringe

will find its target, will remove the cells,

kill them, locate them from A to B,

recruit new cells to fix that tissue,

and have the computation capacity of a real computer.

So how can this be done?

It sounds a bit far-fetched, but let

me explain to you how we can already do that.

And I hope it will be enabled in the very near future.

So if you could look inside that drop of saline magnified

by about 70,000, you will see billions and billions

of tiny objects.

Each of these objects that you see here

is a robot, with chassis and moving parts.

A robot that can be programmed to do amazing things.

These robots are built from DNA.

We use a technique known as DNA origami, to take a piece of DNA

and fold it into the 3D structure

that comprises the robot, the machine.

These robots were born about four years ago in the work

I did with Shawn Douglas at the lab of George Church at Harvard

Medical School.

So each robot has a very specific task

it knows how to do.

It picks up a cargo at point A, and drops the cargo at point B.

Now, in fact, it doesn't actually drop the cargo.

It rather switches the cargo from a concealed state

to an exposed state.

So basically, it switches the cargo from off to on.

And it can go back to off, and it can do this repeatedly.

In the past two years we learned to design

robots that are invisible to the host immune system,

to the mammalian immune system.

We learned how to tune their stability in the blood,

and now it ranges between several minutes

and many, many hours.

So it might seem very simple.

It might seem very limited in its capabilities.

But building on this very basic routine of pick cargo

at A, drop cargo at B, we can build

an astonishing array of tasks.

And what I'd like to do now is to walk together with you

through the tasks required to perform successful surgery.

And let's see how each of these can

be translated into the language that robots understand.

So the first thing is that a surgery takes place

at a specific interface.

That is the interface between the target

issue and a background tissue.

Now, a trained surgeon can manually define this interface

with a precision of about maybe a submillimeter.

But these robots can do tens of thousands of times better.

You can think of them as scalpels

as sharp as a molecule, basically.

So by programming robots to identify both target

and background tissues, we can have them organizing themselves

around that interface.

And now they can expose and activate

a cargo, which in this case is not a drug.

It's an enzyme that can cut tissue components or matrix

components, cut the links between cells,

disintegrating the tissue cell by cell.

The second thing that you have to know

how to do to make a surgery is to move cells

from A to B. You can either, in the most basic sense,

remove your target tissue to the trash, in which case

the trash is point B. Or you can recruit cells

from another place to come and fill in the gap,

and regrow and regenerate the tissue.

So that's also moving cells from A to B.

The robots know how to do this just like ants

do when they carry prey much larger than themselves.

They recognize the specific cells, as you can see here.

They detach them from their surroundings.

Then they can carry specifically those cells, not

the other cells.

They can carry them from point A to point B based

on a combination of molecules defining points A and B,

which is how the robots know to differentiate between points A

and B.

The second thing is, once you have the cell,

you need to know how to reprogram its biology.

You need to tell the cell to either die,

or you can again recruit cells and tell them to regrow,

and fix the tissue and regenerate that tissue.

So we've already shown in numerous examples,

including the one we published two years ago, that when

we load the robots with cancer drugs,

with proteins, with growth factors,

we can tell the cells to either suppress their signaling

or die eventually.

Like you see in the upper figure.

Or we can also tell cells to grow,

as you can see in the example of robots loaded with insulin.

So by the way, when the robots grab their cells,

and now they can engage those cells with selective signalling

molecules, you can have the robot

scan your entire system looking for metastatic cells.

And finding each of these single metastatic cells,

holding them, making sure they don't attach anywhere,

but also now acting on them, killing them,

making sure no cell remains.

So the last thing, in addition to acting on that interface

and on those cells, you need to be able to provide support.

For example, you want to regulate bleeding,

to prevent excessive bleeding from that surgery.

You want to reduce pain.

You want to negate inflammation.

There are many things you need to do that are not directly

related to that incision you're doing right now.

So we already know that these robots

can be programmed to carry out all those tasks.

I want to show you a specific example of robots

that can selectively target and suppress nerve cells,

nerve cells that transduce pain signals.

So the way they do that is they look

for cells that release above a certain threshold

of neurotransmitter.

And only that threshold activates the robots.

And they expose a calcium channel blocker,

which blocks specifically those nerve fibers.

So they can suppress pain deriving

from peripheral procedures.

And eventually, you want to take all those components

and to integrate them such that they are performed

automatically, just as if a computer would control them.

Right?

And that can be specific to something happening

inside a patient right now, and not

something that's constantly-- you

should be able, for example, to stop that procedure whenever

an excessive damage is caused.

So to do that, we designed robots

that can mimic the interactions between components

in a computer processor.

These robots, as you see here, can actually

interact with each other.

It means from point B to these robots

is actually another robot.

It's not a tissue.

So they can communicate and transfer

bits of information, which are also

DNA molecules, from robot to robot.

And these can emulate successfully

logical operators of the basic components of computation

inside living animals.

Those inputs that these robots read

are molecules from the animal.

And the output is written in the form

of drugs acting on the target cells.

So what we do is we combine those robots.

We mix populations of robots such

that we can generate more complex computations.

So this is an example of a half adder which received two input

bits, two input molecules from the animal or from the patient.

And you have two output robots, one representing the carry

bit, one representing the sum bit.

And they can generate a combination of drugs.

And this half adder can be joined to a one-bit adder.

And you can scale that up ad infinitum, basically.

So what we do here is-- this technology

enables us to take it a very complex truth table, which

is basically adapted from an oncology textbook.

And that truth table means, for this disease,

activates this combination.

And what it enables us to do is to take a group of robots,

load them with the entire ensemble of drugs,

and let the robots decide on themselves, based

on what they find in the patient, which combination

to activate.

Moreover, because they sample constantly

the patient's environment, they can

know when to switch to another combination

if the tumor, for example, develops resistance.

So we use logic synthesis to infer the architecture that

generates that truth table.

And this is an example.

So you see a multilayer structure.

And you have a first layer of searching for markers,

searching for disease-specific markers.

And in this example, lung cancer.

And the first layer receives those markers,

generates an output, which is fed also as an input

to a second layer.

Each layer takes about between seven to 10 minutes to process.

And the error is not that high.

The error is in the order of the root sum of squares

of all the errors of individual robots you add in the system.

So eventually that group knows to produce

the exact combination of drugs that an oncologist would also

decide to activate.

Just to wrap this up, if we know all the physical addresses made

by molecules in the patient's body,

we can program those robots to target specific points,

and go to specific locations, produce molecular resolution

incisions, and completely revolutionize

the entire field of surgery as it looks today.

So just imagine-- to sum up-- just

imagine how this could look like,

performing surgery in a drop of saline.

How much this would cost?

Basically nothing.

And most importantly, how can this

become suddenly available to many, many people

who need surgery but can't get it.

Thank you very much.