One of the hardest things about dealing with COVID-19 is knowing what its symptoms are either before, during or after the obvious onset of this illness.
It's been a real slippery thing to try and characterize that way and detect as it goes through its arc.
And yet that's going to be key to managing it spread now and in any future waves.
Professor john Rogers may have some answers for us.
He is the director of Northwestern University's center on bio integrated electronics, a division of Northwestern that I'm pretty sure didn't even exist when I graduated there back in the mid 80s.
John, this sounds like a pretty exciting, innovative part of the school.
What's the short background on it?
Yeah, well, it's kind of a new Institute actually.
It's called the Querrey Simpson Institute for Bioelectronics.
So I spent the first part of my academic career at University of Illinois, Urbana Champaign, got started Bell Labs and then moved to Unifying 2003.
And so I was recruited to Northwestern in 2016.
And part of the recruitment package was to establish this institute to really sort of catalyze efforts at the interface between engineering and medicine course long before this pandemic emerged and our vision has always been to democratize sort of ICU grade.
Healthcare monitoring capabilities so that you can track patients outside of the hospital but with the same kind of ICU grade quality of data streams to enable remote monitoring to improved care in resource constrained, Regions of the globe.
And so we've been working in that space for the last decade and it turns out a lot of those technologies are highly relevant.
COVID-19 and the pandemic were struggling
I find that very concept you just put into a shortened phrase there the idea of ICU grade monitoring on the person wherever they may go to be just full of so many interesting opportunities as we try to.
Get a better handle on how the world's population is doing and I guess you really had it come right to your door.
When COVID-19 came out.
What does this sensor do?
What does it pick up?
First of all?
Yeah, so It's kind of a unique type of platform from the electronics hardware standpoint.
It's a soft and flexible type of device.
I kind of have one here Wearing one right now myself.
It's sort of flexible and soft.
So that's skin compatible, so you can thing of wearing the device on regions of the body other that the wrist, which is where convectional waerables kind of reside and they kind of constrain to And the suprasternal notch, which is where this device is located, this sort of soft tissue location at the base of the neck right where the collar bone sort of joint appears.
And it's a very special location if you're interested in respiratory activity.
And that's key to A lot of the symptoms associated with COVID-19 is cough.
It's shortness of breath fever, of course.
But with a device mounted in this location was something you can't do with a traditional wearable you can provide measurement capability.
Abilities that allow us to determine certain characteristics of underlying physiological processes by measurements of subtle motions of the skin at this part of the body so you can pick up sounds essentially associated with Respiration, sounds and characteristics associated with coughing.
You can actually pick up heart rate and heart rate variability from the pulse little flow of blood through the heart of the artery.
So it's a very special location.
We were interested in that device in the context of stroke, stroke survivors and rehabilitation between measure swallowing and speech, and they have to relearn how to follow and speak, and so our collaborators downtown, we're kind of aware across the medical complex downtown were kind of aware of that platform and reached out to us and asked if we could sort of adapted, modify it, customize it to.
Two COVID symptoms fever, cough and respiratory activity.
And so that really launched us down a very rapid, paced engineering development activity around the software and the hardware to adapt it for those purposes.
And I think we made a lot of great progress.
It's still early.
We're three weeks into this, but We have nearly a terabyte of data.
We're on 25 patients and healthcare workers downtown and it's providing new insights data that was not previously captured in any manner whether whether in the hospital or the home coughing and respiratory sounds.
Yeah, what's so interesting here is you've gone beyond we hear over and over about kind of the same two or three signals being gathered by different wearables and used in different ways but you're picking up something different From somewhere different the Supra sternal notch you're talking about what's inside that you've got an accelerometer.
You talk about listening for cost, but you don't have a microphone.
Is that right?
Yeah, I think you could.
Conceive of using a microphone to measure coughing but I don't think anybody would wear a microphone all day long.
[LAUGH] It's picking up your conversations and all kinds of privacy issues so this device is not based on a microphone.
It's more like a stethoscope, Guess like a digital step, but wireless skin like and it just sits there and records all day long and we capture a lot of data, it gets uploaded to a secure cloud.
And then we put algorithms on top of it to extract these key features coughing respiratory rate, respiratory sounds and we also have a very sensitive temperature sensor and embedded in the device as well.
And so so we capture all this stuff.
So how does the information get off of that into the cloud?
Because I'm always fascinated by the ease of use for the end user if something wants to scale.
Well, yeah, that's a great question.
It's typically a topic that we don't spend a lot of time on as an academic research group.
We're more interested in discovery around the sensing mechanisms and the insights into the data.
But to have an impact to get it really deployed on COVID patients you really have to think about the burden on the patient.
Because these These folks are sick right at the very highest, most serious level right?
And so they don't want to futz around with a piece of gadgetry and go through menus and click on icons and stuff like that.
It's just not possible.
It's also not possible for the for the healthcare workers to futz around with it either because they're busy, right?
They don't have time.
So you We had to spend time on thinking about how to make this as transparent and burden free as possible.
So the device goes on as the sticker goes on.
At the end of the day you pull it off, you drop it onto a charging pad.
As soon as The device sees that it's on a charging pad, it automatically initiates a wireless data transfer to an iPad, and then when it lands on the iPad, the iPad automatically sends it up through a cellular connection to a hypo-compliant cloud Algorithms are automatically applied to that data.
It creates a graphical dashboard that physicians can look at.
And so that's it.
You don't ever have to do anything with the device.
You don't have to mess around with an app of any sort.
When you pull it off the charger it automatically turns on and starts recording.
So then you mount it back up, and that's it.
Okay, so it's real simple, stick it on or charge it.
It's kind of got two modes for the user to deal with.
Now let's go to the Holy Grail here.
You just mentioned once the data is offloaded, it goes into your algorithm.
This is of course, where the magic happens because it seems like at least to a lay person, picking up some relatively simple signals doesn't seem to get you very far unless you have got some magic you can do with those.
What are you doing?
Well, I would say two things.
One is the algorithms that we have right now are just built on deterministic 30 Digital filtering.
So math operations just applied to the data.
We can extract a coughing event.
We can detect those events very, very cleanly.
We can separate them from speech and motion and things like that.
Same thing with respiration.
Same thing with hardware, we can do those classifiers.
And that's what the physicians wanted, and we're responsive to their requests.
This is not a technology push.
It's a clinical poll.
And so that's what they wanted.
And that's what we've delivered.
And I think in the past, they've kind of Qualitatively made assessments of whether a patient is coughing more or less one day to the next, and try to use that in an overall Assessment of Healthcare being able to do in a quantified manner, being able to pick up very early signs of white coughing.
Before maybe a patient is even aware of it becomes very powerful and being able to sort of monitor the progression of a patient through various stages of the disease using these novel biomarkers, coughing, respiratory sounds is, very powerful, but I think the future is to apply artificial intelligence and machine learning algorithms to these same data streams to extract deeper insights into what is the nature of the car, not just whether you coughed or you didn't cough but what sort of cough was it was it a dry cough, was the wet cough the patient swallow after the call?
Interesting Are the costs occurring in coughing fits are they just kind of uniformly distributed?
Are you coughing more at night than you are during the day all this kind of information.
We think could provide additional insights into the basic mechanisms of how the disease is interacting with the body.
And I think as a research tool, it could be important in that context.
That maybe more importantly, you can sort of make quantitative assessments of where a patient is in terms of their progression through the various stages of the disease.
So that's kind of phase two.
Phase One is just give the physicians what they asked for, and that's what we've done.
But then phase two is to really mine this data, right for deeper insights and maybe additional meanings.
So it sounds like you're going down the path of saying we'd start early saying we can detect what we know we're looking for.
And then as we go a little further, we can help find out what we should be looking for.
That's a good way to phrase it.
Now you're also talking about interesting array of patients from one side over here who may have barely any symptoms that are indistinguishable from Allergies today, two people over on the other side here who are coming out of a hellacious bout with COVID-19.
That's a lot of people this seems applicable to.
How do you get to scale what party would take this and run with it?
Do you think as you get down the road.
Yeah, I mean, we're an academic group.
And so we can't scale ourselves.
But we would be very willing to work with outside entities who were interested in scaling if there's the need, right and so we're just responding on, On an ad hoc basis to people who have approached us and asked us about capabilities and I think that may be a good way to start.
We've been funded in the past by BARDA, which is a federal organization that funds this type of work and Maybe in the end of the day they become a vehicle for putting the various parties together if there were a desire and a need and a pull to scale up.
I would say there are probably very few challenges, fundamental challenges to doing that.
I think it's a much simpler device to manufacture than the ventilator for example.
Yeah So the kinda processes are very well aligned with the consumer electronics industry.
So, there are companies out there who know how to do this and we'd be happy to to engage if if there's the need for doing that kinda thing.
So i think there's a clear pathway to scale to very large volumes,if that became,, something that was needed.I think because it's building off of consumer electronics, there's a cost structure there.That probably makes sense.
I don't know that we've done the full bill of materials and you'd have to think about the volumes and so on, but i can't imagine devices Is costing more than maybe 100 bucks.
You know if you're, if you're a scale up, probably less than that, but kind of kind of in that range, so
So here's the big picture question.
You're dealing with one syndrome right now and this has its roots in dealing with recovery from stroke.
So there's a couple of use cases, but as I hear you talking about this And I understand there are other functions you can add to it perhaps a what an oxygen perfusion or blood perfusion.
This becomes something close to the Star Trek tricorder.
But instead of having this big, clunky piece of electronics, there's this very, almost imperceptible thing we could wear down the road a number of years.
Shouldn't we all be wearing a package of as you say, I see you grade medical sensors.
As part of everyday life, isn't that the big goal down the road?
Probably, that's kind of a vision that we've had others as well, you know, over the last decade or so I think it takes on a whole new level of urgency right, given given the In the pandemic and maybe there's a broadening awareness of, you know, developing these kinds of capabilities.
I think it has the potential to really transform the way that we think about health care telemedicine, digitally enabled, you know.
Patient specific insides and at home monitoring rather than episodic measurements when people come into a hospital.
I think the technology is there.
I think we and probably others as well kinda know how to do it.
We're engaged in global Payments of devices of this general class focused on maternal fetal, neonatal and pediatric health just because we think or we used to think that that that population could benefit most strongly.
From a wireless card.
A continuous monitoring capabilities are deployed in 15 countries, five continents, Several places in Africa funded by the Gates Foundation of say the foundation where there are no monitoring technologies at all.
So thinking about an evolution I think you have analogy there as they leapfrog land land lines straight to cellular, maybe they go straight to these wireless devices.
Forget about the old style, wired, ICU, Great monitors, but I think it's a very compelling vision for the future and it's in technology enablement, that's more or less here, the question is just the timescale of deploying it.
I think it could have a lot of utility across a range of conditions.
So let me cut the other way now and play the devil's advocate with an awful lot of the medical and health community will do It's a fair amount of tradition there a fair amount of momentum of doing things a certain way, the way they've been done.
And a lot of them will point to AI in general and you're using something in the AI realm and say this stuff is so far from being able to do what even a in a recent medical school grad can do looking at traditional computer screens and even charts.
What do you say about the current sort of backlash against AI in health, It seems to have cropped up in the last six to nine months?
Well, I think it's a good question.
Our devices don't rely on AI.
I think AI is a value add, a potential value add in the future.
But we work directly with the clinicians, the physicians, we understand that community, we're not trying to push things on them.
We're developing devices that are
Creating data streams that they know are important.
Now what they want to do with that data, they want to put it through a machine learning algorithm to extract insights fine, they can do that.
Or they just want to look at it in the old style method of AI inspection and experience.
That will be their decision.
We're hardware guys.
And so what we want to do is take all the ICU grade monitoring systems, put it into the home in a way that's completely imperceptible physically and not disruptive in any way to natural daily activities.
And then you'd have a lot of options around what you want to do with that data.
We focus on the hardware and clinical grade data.
And I think machine learning is part of that.
But it's not kind of at the core.
It's, it's not a necessary, aspect of the value add I guess.
>.>Okay, so you hand off what your hardware can detect.
And it's up to other parties to decide how low or how high, how manual or how futuristic they want to apply technology to gain insights from it.
Yeah, what we've found, and I think it's a great great point you're making is that, you gotta show up, and engage with the clinicians based on data that they're used to looking at.>>Yeah.>>Data that they've been trained to understand.
You can't show up with a novel matrix, that they've never heard of before.
So we try to reproduce what they're doing today.
But in a totally different format wireless applicable outside of the hospital on and on.
And I think over time the tremendous data flows that will come naturally from that kind of monitoring will open up opportunities for machine learning, but we're not trying to convince anybody to use machine learning.
You don't need to to Realize the benefit of this this kind of technology.
Okay, as we wrap up here, what do you expect is going to happen in the month of May with this project?
we're taping today on the fourth of May.
By the end of this month, what do you think will have been learned, accomplished or move forward?
Well, I think we're going to double up the number of devices that we have deployed is still kind of a small number in the grander scheme of things.
But we'll go from 25 to 50, probably a year in the next couple of weeks.
We're very eager to see how these various metrics, these novel biomarkers that physicians know are important.
How do they evolve as a patient recovers for example or as a patient Begins to recover but then deteriorates.
What are this specific yo insights that we can get around COVID-19 itself?
I think that's something that's going to happen very naturally just over the next few weeks but kind of in parallel with that.
This is the first kind of public you know, announcement of what we're doing and depending on the inbound interest in the folks who want to engage, you know that will determine to a large extent where things go in the future.
Okay, it's a new sensor, it's flexible, it's adhesive and as you can see it's almost imperceptible has its roots in working with stroke patients and Found a whole new role and spotlight here in the emergence and battle against COVID-19.
Professor John Rogers is director of Northwestern University's center on bio integrated electronics.