May 12, 2025 • 20 mins
Implantable medical devices are creating new therapeutic and monitoring solutions for many complex health conditions. However, wireless medical devices are susceptible to malicious attacks. Kaiyuan Yang, associate professor of electrical and computer engineering at Rice University, discusses biomedical security and developing hacker-resistant implants.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:03):
This is the Discovery Files podcastfrom the U.S.
National Science Foundation.
Innovative medical technologies
are transforming how treatmentis approached for modern Americans,
from helping to control seizuresto pacemakers and insulin pumps.
Implantable medical devices are creatingnew therapeutic
and monitoring solutionsfor many complex health conditions.
(00:25):
However, these smart devices
need to record and transmitthe most private data wirelessly,
and that opens up the potentialfor malicious attacks.
We're joined by Kaiyuan Yang,associate professor in the Department
of Electrical and Computerengineering at Rice University.
Among the projects in his Secureand Intelligent Microsystems lab
is one working to protect implantablemedical devices from cyber threats.
(00:48):
Professor Yang,thank you so much for joining me today.
Thanks for inviting me.
So how might hackers or a malicious agent
be able to disrupt a medical devicethat's implanted in someone else?
There are generally two approaches.
So the first one is through cyber attacks.
So you have probably heard aboutphishing attack or password leakage right.
(01:10):
Those are primarily targetinglike a website credentials
or accessed through your mobile phoneor PCs.
Similar attackscan be applied to medical devices
because nowadays most of these medicaldevices have internet access.
They may not be
directly connected to the internet,but it will be through some gateways.
And then somewhere, for example,the patient itself, himself or themself
(01:34):
or the doctors,they may have the logging information
and then attacker could use thiskind of social engineering, right,
like a fake websiteor other ways to steal these passwords.
And then they will bypassall the security protections
in this internet channel and then get direct access to the device.
(01:55):
The other types of cyber attacks, such
as impersonatinginside the within the network, right.
They could impersonateto be something trusted.
And they're interpreted the communicationbetween the legitimate server
and the implant to modifythe message send to the implant,
and that may modifiedoperations of the device.
So broadly speaking,this is the first class that leverages
(02:18):
this existing cyber threatsto hack these devices.
And then another angle is physicalattacks.
One is to interferewith the wireless communication.
Because typically this medical devicethey have special protocols
and physical mechanismsfor various complications.
You can try to, interfere with that, doother things on the communication side.
(02:41):
And also you can use interferenceto change the sensor data.
There has been reported demonstrationsof using ultrasound
or other RF interference to modifythe data
being collected by the sensor,and then needed to make around decisions.
In the wireless spectrum.
Thinking about radio broadcasts.
And now with our phones like 5G, differentkind of bandwidths that happen.
(03:04):
Do biomedical devices need to workin a specific kind of frequency area?
It's a great question.
There is not a consensus on that yet.
Most of the commercially available devicethat they tend to be on the larger size.
They typically adopt commercial radios,
which means they're at the gigahertzrange, or at least several hundreds
(03:27):
of megahertz range that allows themto radiate to a further distance.
But then the challenge withthis is our signal.
It's being absorbed by the human tissue.
And the higher frequency you go,the more absorption, there will be.
And that will cause heatingand safety problem
and ultimately limithow much energy you can send out.
(03:47):
So for more recent research,especially like, like ours, that
we try to build miliimeter scaleimplants, in these cases
we tend to use lower frequency mechanismsto send the various signal.
And we are not only usingtraditional electromagnetic waves,
we also use acoustic wavesor other mechanisms to reduce
(04:09):
the size of the transmitter and receiversthat's required for the implant.
What kind of interferenceor other RF signals might get in the way
of your millimeter scale devices?
So actually the benefitsof going to these lower frequency twofold.
So on one handis to give you better safety limits
(04:29):
because at lower frequency,the amount of energy you can send out
can transmit through the body,which will be much higher.
So it's safer.
And then on the other hand,because this will sit
at a very different frequencyfrom the more traditional like,
radio frequency bands,it will suffer from less interference.
So we will always have a, low pass filter
(04:51):
to only focus on the frequencythat's of interest to the device.
So that will improve the resiliencetowards this, interference.
Very cool.
What are some of the techniquesyour lab is exploring
to make these kind of devices more secure?
So, as I mentioned,there are two sides of the challenge.
(05:15):
Ideally,your solution is to handle both so well.
For our most recent workwe are trying to provide
two factorauthentication to these spinal implants.
So the goal is very simple.
And nowadays you want to loginto your bank account right.
You enter your username passwordand then they don't directly accept you.
(05:35):
They send you a message,a random number to your phone
to verify you are aware of this.
Log in and then you have to approve that.
Or you enter that passcode to the websiteand then they log you in.
So this is called two factorauthentication.
If you want to have the same abilityfor these biomedical implants.
So that when the following credentials
(05:57):
from the doctor, when the patientgets lost that tag her own.
To be able to easily
get into the device and modifyas parameters or steal the data.
The challenge is achievingthese two factor authentication is that
think of our normal process
where you have a cell for the cell phone,respond data to this user,
and then the cell phone has a screenso it can receive this message.
(06:20):
And then you have your keyboardfor your computer
which you can enter this passcoderight before a vocal implant.
We do not have this monitorand this keyboard.
So we need to have a way to reliably send
a fresh key,fresh random number into the implant.
And the implantmust be able to recognize that.
And I'll give you the order to do thiswithin a millimeter scale device.
(06:44):
We must have a good wayto enter this number.
And what we realize isour devices are always powered
by external power transmitterbecause you want to make it battery free,
or it can work with a very small batterythat is recharged constantly.
In this sense,when you move the external transmitter,
(07:05):
the power transfer efficiency will change.
So the amount of power you receivewill be dependent on the alignment
between the external transmitterand the implanted device.
So we utilize this as a mechanism, to enter the passcode.
So we call this a mechanical input.
And the good thing forthis is it is almost a free
(07:28):
add up to the existing powertransfer scheme.
It does not require any additional sensorsor additional input ports.
It just use the existing power transferlink and the power receiving circuits.
We are able to detectthe user enter pattern from outside.
And this gives us the keyboardto enter this new, random number.
(07:49):
So if there was like an emergencyaccess case where like a paramedic
needed to get in, how would theyget through that two factor.
When the first question.
That's one of the unique challengesfor doing the modification
in medical devices, because we have toconsider emergency scenario.
And we must assumethat in the emergency scenario,
(08:12):
the first respondersmay not have access to your password.
So we have to disable or modify the input.
So what we do with that is our protocol.
We build a complete full stacksecurity protocol for this.
We do have like emergency modeand emergency mode.
We also use this secondfactor mechanical input.
(08:33):
But we will bypass the first factorwhich is the password.
And the way to make surethe second factor is secure.
It's we utilize the distance bondfor our communication skills.
So because our communicationis at a low frequency,
and then we have a very clearunderstanding of how far it can transmit.
And we know exactly like giventhis amount of power
(08:56):
for transmission,how long the communication will be.
So we can guarantee thatonly someone who has the receiver
that's placed, for example,ten centimeter near the implant
will be able to receivea fresh random number.
And then the person who receive thisrandom number can move
your wearable transmitterto enter the passcode.
(09:19):
So the assumption here isthe patient is supposed to be aware
of these operations because the person whomanipulate this, the first need to place
the receiver very close to your bodyto receive the fresh code.
And then they have to move your powertransmitter to enter the passcode.
We designed this passcodeto be long enough so that someone who
(09:40):
maybe you just, catch you temporarily.
You know, a bus or railway won't haveenough time to do all these things.
But also, it's not too longso that it's, compromise the convenience.
So we try to strike a balancein choosing a proper advance of the past.
In your secure
and intelligent micro systems lab,you're working on tiny scale things.
(10:03):
Are there problems gettingsome of these features down to that size?
Yes, definitely.
That's actually one of the keyresearch slots for our network.
We are trying to build security solutionsfor this ultra small and all channel
power systems, not only for biomedical,
but also for a generic Internet of Thingsdevices.
The challenge here isyou have very limited hardware resource,
(10:27):
which means you can only havea very small chip or some,
limited number of off chip componentsfor your security function.
The energy budget is extremely restricted.
Traditionally,if you do all this security scheme,
you have all the open source librariesto do them in software.
Right? Normally you think about security.
Is the source for a problem,
but then that relieson a powerful processor that can rule out
(10:49):
whatever protectionor algorithm encryptions you want to do.
But in these extremely scaled,
resource constrained devices,you don't have that access.
So what we do iswe use specialized hardware
and special chip designsto support security functions
that are at a similar strengthas the conventional approach,
(11:12):
but only take orders of magnitudesmaller area in power.
And these include componentsfor, random number generation.
And as I mentioned, we need fresh keysfor all of the security product
protocol stuff.
So we have built in low powerrandom number generator.
We have phase coming out kernel function,which provides
(11:32):
a device uniqueID and secret keys for each implant.
And then we have low powersecurity encryption engines
so we can perform, symmetric keyAES encryptions
and also, hashing operationswithin this millimeter scale implant.
So it's really we put a lot of our effortsin these individual components.
(11:55):
And then we integrate them togetherinto this, bio implant platform,
which enables this, two factorauthentication capability
for this implant.
As you're thinkingabout rolling these out in the future,
are you consideringsome of the technology impacts as like,
hackers get used to using supercomputersand AI kind of techniques?
(12:19):
Are you trying to future proofthe security functions in that way?
Yeah, that's another great question.
On one hand the cryptography world
people are well aware of the challengefrom quantum computers,
which are known to be ableto break all the existing ciphers.
So there are post-quantumcryptographic algorithms.
(12:42):
They have recently been,approved by NIST.
So they're, like, recommended algorithmsfor post-quantum cryptography.
And I was also looking into that.
So we are buildingthis, dedicated hardware that can support
those kind of post-quantum algorithms,within a small power budget.
But for this particular,
(13:04):
this secure implants work,we are only using symmetric key ciphers,
which has less vulnerabilityto supercomputers.
So most of the supercomputersand quantum computing attacks
are targeting public key cryptography.
And the our scheme,we are not using public key cryptography
because that is too expensive in termsof power and hardware requirements.
(13:27):
Yeah.
So we are taking alternative routesbased on private key.
So yeah, from that sense,I think that's probably nice of a concern.
And then the other side,that technical advancements
may impact this security scheme isthere could be more powerful
wireless receivers and sensorsso that that may break our distance bound.
(13:50):
So, you know,if the attacker is at home, right.
Under these times,they can still steal the information.
In that case, as I mentioned, we havea clear understanding of the signal
to noise ratio versus the amount of powerwe put out of the inplant.
So we could dynamically adjustthis power level
to make sure that even if your,
(14:11):
receiversensitivity is improved by ten DB,
we can still guarantee thatat a certain distance, you won't be able
to get sufficient informationto decode a random code.
So I would say in our current scheme,it has some kind of future proof.
But we have to seewhat happens in the future.
And security is always kind of a catand rat problem.
(14:32):
So there's no perfect security.
And you have to have an understandingof the capability of the attacker
and then try to make it difficultor impossible
for attackers with the given capability.
From this point in time, kind ofwhat is your path forward
to getting thisinto commercial applications,
or to where the average personmight have it in their personal device?
(14:56):
We have been working on this,
bioimplants for probably 7 to 8 years.
It's mostly a collaboration withmy colleague at Rice that Jacob Robinson.
So we have been developing thisMagnetoelectric based bioimplant platform.
We actually already started a companythat tried to commercialize this.
(15:16):
And we are going through the processof getting it prepared
for FDA approvaland to do first in human studies.
So we have already built several, versions of the prototyping device.
We have performed animal studies,and they have shown very good
performance and efficacyin animal studies.
So that's why we are building this,transferring
(15:38):
this to to a more commercial setup.
And hopefully that will resultin something that will help the patients.
The first therapeutic targets for our,commercial device, it's, mental health.
So we try to use, you know, electricalstimulation to help alleviate depression.
Interesting.
(15:59):
Professor Yang,
can you tell us a little bit about how NSFsupport has impacted your work?
Yeah.
NSF support has been tremendousfor our research.
And for different aspectsfor this, medical research
over, one of our initial grants with DocRobinson
was through the ascend program,and there the engineering directory.
(16:21):
It help us build the basic platformfor this file, medical device.
So I support us to viewthe various applications of various
power and networking capabilitiesto these implants.
And then after that,I got my, career awards,
which funds the effortsto make it more secure and more reliable.
So building on top of the initial platformand based on these technologies
(16:44):
we built and supported by these twogrants, we are also getting other,
federal agenciesto support more application
specific usageand developments of this work.
So now we have, otherfor more like a pain relief for,
cardiac applicationsand also for, neuro disorders.
So it's really help us kickstartall these efforts,
(17:07):
and now we have multiple targetapplications that's on their development.
Thinking about those target applications,I guess
for my last question, I want to ask youabout what's coming up next in your work.
Where do you see this workgoing in the next few years?
Let's say on the clinical side,
we are tryingto make our very basic function because
(17:28):
now there is not a single commerciallyavailable millimeter scale impact.
So we want to be one of the firstto get that really deliver to the patient.
So to do thatwe need to make our power transfer
communication, sensing and stimulationthese four basic functions
to be as reliable as possible, rightas you can imagine,
(17:48):
we want these thingsto be super reliable and trusted.
And we also want to better understandthe clinical impacts
that we can bring with such a new tool.
And then on the research side,we are trying to build
build more advanced functionsinto these devices,
such as we want to have closedloop neuromodulation,
(18:09):
which means we want the device to be ableto sense the signal in real time
and then make decisionsusing artificial intelligence
or other decision making capabilities,and then determine
the stimulation patterns, the electrical
or optical stimulationsthat will best help the patient.
This is not possible today.
Today, the typical flow isthe doctor needs to adjust this manually,
(18:33):
and they don't havea very direct indicator to tell them
whether this is working or not.
So it's a very lengthy process, right?
The patient has to go back,return to the clinic
many times to adjust the parameters,and maybe they find a good parameter.
But after a year the parameter will changebecause the idea of the patient
will change over time.
So I think one of the Holy grail
(18:55):
for this iswe want to achieve the closed loop.
This will require a new understandingof maybe neuroscience, the clinical side,
new algorithms, and alsonew hardware support as we want to add
AI or other more advancedsignal processing to the device.
You facethe same power problem as security.
So we have to build very efficienthardware
(19:18):
that can support these AI models
and to support its closed loop operation.
So our device will evolveinto different form factors.
Some of them,
we try to make it as small as possible,and then they have a few actual channels
they can record from a few, channels,and they can stimulate a few channels.
(19:39):
This, for more for therapeutic effects.
And then we can also scale upthe platform to support hundreds
or even thousands of channelsfor recording stimulation
that will become the so-called bridgemachine interface.
And that opens up new waysto interact with the human brain.
And to have a lot of more futuristicapplications, like,
(20:02):
prosthetics or robotic armsthat's be controlled by demand, companies
like Neuralink, have makea lot of progress towards that goal.
But in terms of the hardware platformand the technology foundation,
I think there's a lot of researchthat needs to be done
to really support that vision.
So we are also working towards that goal.
(20:22):
Special thanks to Kaiyuan Yang.
For the Discovery Files, I'm Nate Pottker.
You can watch video versions
of these conversations on our YouTubechannel by searching @NSFscience.
Please subscribe wherever you get podcastsand if you like our program, share
with a friendand consider leaving a review.
Discover how the U.S.
National Science Foundationis advancing research at NSF..gov.
This is the Discovery Files podcastfrom the U.S.
National Science Foundation.
Innovative medical technologies
are transforming how treatmentis approached for modern Americans,
from helping to control seizuresto pacemakers and insulin pumps.
Implantable medical devices are creatingnew therapeutic
and monitoring solutionsfor many complex health conditions.
(00:25):
However, these smart devices
need to record and transmitthe most private data wirelessly,
and that opens up the potentialfor malicious attacks.
We're joined by Kaiyuan Yang,associate professor in the Department
of Electrical and Computerengineering at Rice University.
Among the projects in his Secureand Intelligent Microsystems lab
is one working to protect implantablemedical devices from cyber threats.
(00:48):
Professor Yang,thank you so much for joining me today.
Thanks for inviting me.
So how might hackers or a malicious agent
be able to disrupt a medical devicethat's implanted in someone else?
There are generally two approaches.
So the first one is through cyber attacks.
So you have probably heard aboutphishing attack or password leakage right.
(01:10):
Those are primarily targetinglike a website credentials
or accessed through your mobile phoneor PCs.
Similar attackscan be applied to medical devices
because nowadays most of these medicaldevices have internet access.
They may not be
directly connected to the internet,but it will be through some gateways.
And then somewhere, for example,the patient itself, himself or themself
(01:34):
or the doctors,they may have the logging information
and then attacker could use thiskind of social engineering, right,
like a fake websiteor other ways to steal these passwords.
And then they will bypassall the security protections
in this internet channel and then get direct access to the device.
(01:55):
The other types of cyber attacks, such
as impersonatinginside the within the network, right.
They could impersonateto be something trusted.
And they're interpreted the communicationbetween the legitimate server
and the implant to modifythe message send to the implant,
and that may modifiedoperations of the device.
So broadly speaking,this is the first class that leverages
(02:18):
this existing cyber threatsto hack these devices.
And then another angle is physicalattacks.
One is to interferewith the wireless communication.
Because typically this medical devicethey have special protocols
and physical mechanismsfor various complications.
You can try to, interfere with that, doother things on the communication side.
(02:41):
And also you can use interferenceto change the sensor data.
There has been reported demonstrationsof using ultrasound
or other RF interference to modifythe data
being collected by the sensor,and then needed to make around decisions.
In the wireless spectrum.
Thinking about radio broadcasts.
And now with our phones like 5G, differentkind of bandwidths that happen.
(03:04):
Do biomedical devices need to workin a specific kind of frequency area?
It's a great question.
There is not a consensus on that yet.
Most of the commercially available devicethat they tend to be on the larger size.
They typically adopt commercial radios,
which means they're at the gigahertzrange, or at least several hundreds
(03:27):
of megahertz range that allows themto radiate to a further distance.
But then the challenge withthis is our signal.
It's being absorbed by the human tissue.
And the higher frequency you go,the more absorption, there will be.
And that will cause heatingand safety problem
and ultimately limithow much energy you can send out.
(03:47):
So for more recent research,especially like, like ours, that
we try to build miliimeter scaleimplants, in these cases
we tend to use lower frequency mechanismsto send the various signal.
And we are not only usingtraditional electromagnetic waves,
we also use acoustic wavesor other mechanisms to reduce
(04:09):
the size of the transmitter and receiversthat's required for the implant.
What kind of interferenceor other RF signals might get in the way
of your millimeter scale devices?
So actually the benefitsof going to these lower frequency twofold.
So on one handis to give you better safety limits
(04:29):
because at lower frequency,the amount of energy you can send out
can transmit through the body,which will be much higher.
So it's safer.
And then on the other hand,because this will sit
at a very different frequencyfrom the more traditional like,
radio frequency bands,it will suffer from less interference.
So we will always have a, low pass filter
(04:51):
to only focus on the frequencythat's of interest to the device.
So that will improve the resiliencetowards this, interference.
Very cool.
What are some of the techniquesyour lab is exploring
to make these kind of devices more secure?
So, as I mentioned,there are two sides of the challenge.
(05:15):
Ideally,your solution is to handle both so well.
For our most recent workwe are trying to provide
two factorauthentication to these spinal implants.
So the goal is very simple.
And nowadays you want to loginto your bank account right.
You enter your username passwordand then they don't directly accept you.
(05:35):
They send you a message,a random number to your phone
to verify you are aware of this.
Log in and then you have to approve that.
Or you enter that passcode to the websiteand then they log you in.
So this is called two factorauthentication.
If you want to have the same abilityfor these biomedical implants.
So that when the following credentials
(05:57):
from the doctor, when the patientgets lost that tag her own.
To be able to easily
get into the device and modifyas parameters or steal the data.
The challenge is achievingthese two factor authentication is that
think of our normal process
where you have a cell for the cell phone,respond data to this user,
and then the cell phone has a screenso it can receive this message.
(06:20):
And then you have your keyboardfor your computer
which you can enter this passcoderight before a vocal implant.
We do not have this monitorand this keyboard.
So we need to have a way to reliably send
a fresh key,fresh random number into the implant.
And the implantmust be able to recognize that.
And I'll give you the order to do thiswithin a millimeter scale device.
(06:44):
We must have a good wayto enter this number.
And what we realize isour devices are always powered
by external power transmitterbecause you want to make it battery free,
or it can work with a very small batterythat is recharged constantly.
In this sense,when you move the external transmitter,
(07:05):
the power transfer efficiency will change.
So the amount of power you receivewill be dependent on the alignment
between the external transmitterand the implanted device.
So we utilize this as a mechanism, to enter the passcode.
So we call this a mechanical input.
And the good thing forthis is it is almost a free
(07:28):
add up to the existing powertransfer scheme.
It does not require any additional sensorsor additional input ports.
It just use the existing power transferlink and the power receiving circuits.
We are able to detectthe user enter pattern from outside.
And this gives us the keyboardto enter this new, random number.
(07:49):
So if there was like an emergencyaccess case where like a paramedic
needed to get in, how would theyget through that two factor.
When the first question.
That's one of the unique challengesfor doing the modification
in medical devices, because we have toconsider emergency scenario.
And we must assumethat in the emergency scenario,
(08:12):
the first respondersmay not have access to your password.
So we have to disable or modify the input.
So what we do with that is our protocol.
We build a complete full stacksecurity protocol for this.
We do have like emergency modeand emergency mode.
We also use this secondfactor mechanical input.
(08:33):
But we will bypass the first factorwhich is the password.
And the way to make surethe second factor is secure.
It's we utilize the distance bondfor our communication skills.
So because our communicationis at a low frequency,
and then we have a very clearunderstanding of how far it can transmit.
And we know exactly like giventhis amount of power
(08:56):
for transmission,how long the communication will be.
So we can guarantee thatonly someone who has the receiver
that's placed, for example,ten centimeter near the implant
will be able to receivea fresh random number.
And then the person who receive thisrandom number can move
your wearable transmitterto enter the passcode.
(09:19):
So the assumption here isthe patient is supposed to be aware
of these operations because the person whomanipulate this, the first need to place
the receiver very close to your bodyto receive the fresh code.
And then they have to move your powertransmitter to enter the passcode.
We designed this passcodeto be long enough so that someone who
(09:40):
maybe you just, catch you temporarily.
You know, a bus or railway won't haveenough time to do all these things.
But also, it's not too longso that it's, compromise the convenience.
So we try to strike a balancein choosing a proper advance of the past.
In your secure
and intelligent micro systems lab,you're working on tiny scale things.
(10:03):
Are there problems gettingsome of these features down to that size?
Yes, definitely.
That's actually one of the keyresearch slots for our network.
We are trying to build security solutionsfor this ultra small and all channel
power systems, not only for biomedical,
but also for a generic Internet of Thingsdevices.
The challenge here isyou have very limited hardware resource,
(10:27):
which means you can only havea very small chip or some,
limited number of off chip componentsfor your security function.
The energy budget is extremely restricted.
Traditionally,if you do all this security scheme,
you have all the open source librariesto do them in software.
Right? Normally you think about security.
Is the source for a problem,
but then that relieson a powerful processor that can rule out
(10:49):
whatever protectionor algorithm encryptions you want to do.
But in these extremely scaled,
resource constrained devices,you don't have that access.
So what we do iswe use specialized hardware
and special chip designsto support security functions
that are at a similar strengthas the conventional approach,
(11:12):
but only take orders of magnitudesmaller area in power.
And these include componentsfor, random number generation.
And as I mentioned, we need fresh keysfor all of the security product
protocol stuff.
So we have built in low powerrandom number generator.
We have phase coming out kernel function,which provides
(11:32):
a device uniqueID and secret keys for each implant.
And then we have low powersecurity encryption engines
so we can perform, symmetric keyAES encryptions
and also, hashing operationswithin this millimeter scale implant.
So it's really we put a lot of our effortsin these individual components.
(11:55):
And then we integrate them togetherinto this, bio implant platform,
which enables this, two factorauthentication capability
for this implant.
As you're thinkingabout rolling these out in the future,
are you consideringsome of the technology impacts as like,
hackers get used to using supercomputersand AI kind of techniques?
(12:19):
Are you trying to future proofthe security functions in that way?
Yeah, that's another great question.
On one hand the cryptography world
people are well aware of the challengefrom quantum computers,
which are known to be ableto break all the existing ciphers.
So there are post-quantumcryptographic algorithms.
(12:42):
They have recently been,approved by NIST.
So they're, like, recommended algorithmsfor post-quantum cryptography.
And I was also looking into that.
So we are buildingthis, dedicated hardware that can support
those kind of post-quantum algorithms,within a small power budget.
But for this particular,
(13:04):
this secure implants work,we are only using symmetric key ciphers,
which has less vulnerabilityto supercomputers.
So most of the supercomputersand quantum computing attacks
are targeting public key cryptography.
And the our scheme,we are not using public key cryptography
because that is too expensive in termsof power and hardware requirements.
(13:27):
Yeah.
So we are taking alternative routesbased on private key.
So yeah, from that sense,I think that's probably nice of a concern.
And then the other side,that technical advancements
may impact this security scheme isthere could be more powerful
wireless receivers and sensorsso that that may break our distance bound.
(13:50):
So, you know,if the attacker is at home, right.
Under these times,they can still steal the information.
In that case, as I mentioned, we havea clear understanding of the signal
to noise ratio versus the amount of powerwe put out of the inplant.
So we could dynamically adjustthis power level
to make sure that even if your,
(14:11):
receiversensitivity is improved by ten DB,
we can still guarantee thatat a certain distance, you won't be able
to get sufficient informationto decode a random code.
So I would say in our current scheme,it has some kind of future proof.
But we have to seewhat happens in the future.
And security is always kind of a catand rat problem.
(14:32):
So there's no perfect security.
And you have to have an understandingof the capability of the attacker
and then try to make it difficultor impossible
for attackers with the given capability.
From this point in time, kind ofwhat is your path forward
to getting thisinto commercial applications,
or to where the average personmight have it in their personal device?
(14:56):
We have been working on this,
bioimplants for probably 7 to 8 years.
It's mostly a collaboration withmy colleague at Rice that Jacob Robinson.
So we have been developing thisMagnetoelectric based bioimplant platform.
We actually already started a companythat tried to commercialize this.
(15:16):
And we are going through the processof getting it prepared
for FDA approvaland to do first in human studies.
So we have already built several, versions of the prototyping device.
We have performed animal studies,and they have shown very good
performance and efficacyin animal studies.
So that's why we are building this,transferring
(15:38):
this to to a more commercial setup.
And hopefully that will resultin something that will help the patients.
The first therapeutic targets for our,commercial device, it's, mental health.
So we try to use, you know, electricalstimulation to help alleviate depression.
Interesting.
(15:59):
Professor Yang,
can you tell us a little bit about how NSFsupport has impacted your work?
Yeah.
NSF support has been tremendousfor our research.
And for different aspectsfor this, medical research
over, one of our initial grants with DocRobinson
was through the ascend program,and there the engineering directory.
(16:21):
It help us build the basic platformfor this file, medical device.
So I support us to viewthe various applications of various
power and networking capabilitiesto these implants.
And then after that,I got my, career awards,
which funds the effortsto make it more secure and more reliable.
So building on top of the initial platformand based on these technologies
(16:44):
we built and supported by these twogrants, we are also getting other,
federal agenciesto support more application
specific usageand developments of this work.
So now we have, otherfor more like a pain relief for,
cardiac applicationsand also for, neuro disorders.
So it's really help us kickstartall these efforts,
(17:07):
and now we have multiple targetapplications that's on their development.
Thinking about those target applications,I guess
for my last question, I want to ask youabout what's coming up next in your work.
Where do you see this workgoing in the next few years?
Let's say on the clinical side,
we are tryingto make our very basic function because
(17:28):
now there is not a single commerciallyavailable millimeter scale impact.
So we want to be one of the firstto get that really deliver to the patient.
So to do thatwe need to make our power transfer
communication, sensing and stimulationthese four basic functions
to be as reliable as possible, rightas you can imagine,
(17:48):
we want these thingsto be super reliable and trusted.
And we also want to better understandthe clinical impacts
that we can bring with such a new tool.
And then on the research side,we are trying to build
build more advanced functionsinto these devices,
such as we want to have closedloop neuromodulation,
(18:09):
which means we want the device to be ableto sense the signal in real time
and then make decisionsusing artificial intelligence
or other decision making capabilities,and then determine
the stimulation patterns, the electrical
or optical stimulationsthat will best help the patient.
This is not possible today.
Today, the typical flow isthe doctor needs to adjust this manually,
(18:33):
and they don't havea very direct indicator to tell them
whether this is working or not.
So it's a very lengthy process, right?
The patient has to go back,return to the clinic
many times to adjust the parameters,and maybe they find a good parameter.
But after a year the parameter will changebecause the idea of the patient
will change over time.
So I think one of the Holy grail
(18:55):
for this iswe want to achieve the closed loop.
This will require a new understandingof maybe neuroscience, the clinical side,
new algorithms, and alsonew hardware support as we want to add
AI or other more advancedsignal processing to the device.
You facethe same power problem as security.
So we have to build very efficienthardware
(19:18):
that can support these AI models
and to support its closed loop operation.
So our device will evolveinto different form factors.
Some of them,
we try to make it as small as possible,and then they have a few actual channels
they can record from a few, channels,and they can stimulate a few channels.
(19:39):
This, for more for therapeutic effects.
And then we can also scale upthe platform to support hundreds
or even thousands of channelsfor recording stimulation
that will become the so-called bridgemachine interface.
And that opens up new waysto interact with the human brain.
And to have a lot of more futuristicapplications, like,
(20:02):
prosthetics or robotic armsthat's be controlled by demand, companies
like Neuralink, have makea lot of progress towards that goal.
But in terms of the hardware platformand the technology foundation,
I think there's a lot of researchthat needs to be done
to really support that vision.
So we are also working towards that goal.
(20:22):
Special thanks to Kaiyuan Yang.
For the Discovery Files, I'm Nate Pottker.
You can watch video versions
of these conversations on our YouTubechannel by searching @NSFscience.
Please subscribe wherever you get podcastsand if you like our program, share
with a friendand consider leaving a review.
Discover how the U.S.
National Science Foundationis advancing research at NSF..gov.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy And Charlamagne Tha God!
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.