Digital Technologies to Empower Individuals
Dr. Anderson is the chief digital health physician here at MITRE, where he leads research and development efforts across major strategic initiatives and digital health, including several with industry and the U.S. government during the COVID-19 pandemic. Dr. Anderson is responsible for co-founding the VCI coalition of more than 475 organizations working towards interoperable and verifiable clinical information, leading MITRE's development of machine learning techniques to leverage real-world health data, and linking genomic and clinical data to provide insight into the efficacy of therapeutics and vaccines. So over to you, Dr. Anderson.
- Thanks, Dawn, and welcome to our third and final panel for the evening. We have a really incredible set of panelists for you today. They are world renowned experts in digital health, public policy, and artificial intelligence. Earlier this week, we recorded our interview with our first panelist, Dr. Robert Wachter. We have the pleasure and honor of having Dr. Robert Wachter as one of our panelists. Dr. Wachter is a professor
and chair of the department of medicine at the university of California, San Francisco. In 2021 to 22, the department was ranked the best internal medicine department in the nation by U.S. News & World Report. Dr. Wachter is also the author of 300 articles and six books.
He coined the term hospitalist in 1996, and he's often considered the father of the hospitalist field, the fastest growing medical specialty in the U.S. history. He's the past president of the Society of Hospital Medicine, the past chair of the American Board of Internal Medicine. He's written two books on safety and quality, including "Understanding Patient Safety", the world's top selling safety primer. His 2015 book, my favorite, "The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine's Computer Age" was a New York times bestseller. He's received several honors for his work.
In 2004, he received the John M. Eisenberg Award, the nation's top to honor in patient safety. 13 times, Modern Healthcare magazine has ranked him as one of the most 50 influential physician executives in the United States. He was number one on the list in 2015. He's a master of the American College of Physicians and elected member of the National Academy of Medicine. His tweets on COVID-19 have been followed by over 180,000 people, retweeted over 200 million times.
He serves as a trusted source of information on clinical public health, and policy issues surrounding this pandemic. Dr. Wachter, thank you for joining us. It's a pleasure to have you here. - Thank you, Brian. It's a pleasure. I'm glad my mother's a rendition of the bio made it through.
(Dr. Brian laughs) - Well, so let's go ahead and get started. One of the areas that I know that you've done a lot of leading thought and comments on is around how can health systems engage the new kinds of patients that are enabled with new tools, a new kind of way of engaging with their providers that extends beyond the 15 minute episodic visit today? - Yeah, I think we're entering a very exciting era and I think there are risks, but I think at the end of the day, it's going to be good for patients and actually good for health systems. I think the key is to understand that for the foreseeable future not many patients are going to live autonomous digital lives, and I think the analogies could get us in trouble here. We don't use travel agents very much anymore. Many people don't use accountants to do their taxes very much anymore.
And you might say, all right, that's the way healthcare is going. Patients will manage themselves virtually, they'll be collecting data, and this magical AI will tell them how to manage their blood pressure, manage their inflammatory bowel disease. I think that digital in healthcare is somewhat different.
I think it's much more likely to be an adjunct to the healthcare system, which means that we've got to sort out sort of what is the relationship between the patient, the tools and apps they're using at home, the data that's being collected, and the enterprise system that has to be there as part of an escalation strategy. So I do believe there will be kind of simple things and we're already seeing this. Patients better able to manage their diabetes, for example, using a digital glucometer, an AI that gives them some advice that helps them manage their insulin. But I think for most patients it's going to look like there's some basic things that happen through digital and virtual, but a lot of it is via information that's collected and then goes to a doctor, goes to a healthcare system. And so we've got to figure this out and here's my worry, Brian.
I have 300 primary care docs who work in my department. I can say to them, there's this new cool digital stuff coming out, all your patients are gonna be wired. Their scales are gonna be wired. Their toilets are gonna be wired, so it can measure their urine electrolytes. And all of that data is gonna stream to you every day automatically in real time, so you have all this rich data on the 1800 patients you're following. I can tell you, every single one of them will quit by five o'clock this afternoon, so we have to think about this in a very new way.
What does the health care system look like that can manage this data and handle it effectively? Now, you could argue that you're gonna need new healthcare systems that were built for that purpose. And I think there is going to be some of that particularly in primary care. But I do think we've got to give a lot more thought to the sort of new layer that we have to build in the healthcare system that oversees this digital data flow can escalate to something that looks still fairly traditional. It's a doctor, it's a hospital. - I love the concept of this remote wellness paradigm that we're moving towards.
What does that middle layer look like in terms of how it would empower providers and patients for that matter, and not overload either one of them? - Well, John Halamka I think coined the term, "Care traffic controllers", and I think it's sort of looks like that. I think it is that they are collecting data every day, maybe because they have a new complaint and they go online and describe it somehow. Maybe it's through sort of automated data flow when they step on their scale it's digital, or they check their glucose it's digital, or their oxygen level. It goes to some combination of people and digital competencies.
And then some magic has to happen where for the basics and when patients are doing fine, they're sort of within the guard rails, the system basically says, attaboy, attagirl, you're doing great. Your weight has stayed stable. Your glucose is good. And no person needs to touch that. But the strategy of what happens when that's not true has to be really nuanced and really sophisticated.
It may be that it's not true that the patient's weight is going up. A nutritionist needs to sort of chime in. Sometimes it's gonna be a generalist doctor, sometimes it's gonna be a specialist.
The system has to be flexible enough that it's sort of understands what the issue is and what the right kind of combination of person and digital is necessary to intervene in the way that's most productive. So that is really hard, figuring out how to make that work in a world where you're gonna be getting a bazillion signals from patients. Most of them are gonna be false positives. Think about the alert fatigue that we have in the intensive care unit, where it's all industrial grade signals and it's all a professional environment that we can manage ourselves. What happens in the environment where the patient forgot to take the sensor off when he or she went into the shower? So separating signal from noise, building an escalation strategy that works, it's really gonna be challenging work to get it right and we're gonna get it wrong in a lot of ways, but one key point is if it's mainlined into the healthcare system and gets right to your doctor, the system will implode. The system was not built for that.
The primary care doctors are already living on the precipice. Many of them are all really massively unhappy, because we've already enabled 24, seven, 365 connectivity through digital means. And so patients are now reading their notes and they're texting their doctors in the middle of the night and results are coming back 24 seven. And we just sort of expected that the doctors can manage that.
So it can't be that the answer is it'll go to your doctor and your doctor will figure it out. It has to be that digital can handle a whole bunch of it, but it also has to be that the doctors, or nurses, or nutritionists, or physical therapists are engaged at the appropriate time in the right way. And one more thing which I think is worth paying attention to as we move toward more autonomous systems the stakes grow and the tolerance that we will all have for bad mistakes is going to be relatively low. And how do we know that? Well, when we see an autonomous car have an accident and someone gets killed, there's really no one who is able to say convincingly, well, when you look at a trillion auto rides, it turns out the autonomous car on average is safer, 'cause they are not texting while they're driving, they're not fatigued, they're not drunk. Yeah, but that error made by the autonomous system where human was involved is very visible, will create litigation. And so there is still is gonna have to be a fair amount of human oversight built in these systems if they're gonna work effectively.
That is a pretty heavy lift. - What sort of guidance or thoughts do you have in terms of as the digital health innovation in different kinds of tools and apps progresses and evolves, how can healthcare systems set up the kind of infrastructure so that when those next tools are enabled to engage with their patients do so in an integrated seamless way while also setting up the provider for success as well? - Yeah, I think that probably the most important thing is for large healthcare systems that can pull this off. It's gonna be important to think about this as a really specific and complicated business problem to solve. And I'll tell you what's happened at UCF, which has sort of informed my thinking about this.
In the old days, our digital system was really the group of people on the clinical side. It was the chief medical information officer and his legions of people. And their job was to put our EHR in and manage it once it was in. And we have Epic, like most of the large academic health systems. And so their job was really to install Epic and then to think about new features, and the upgrades, and listen to the providers moan about this and that, and deal with it and all that kind of stuff.
That group of people has to be laser focused on that set of tasks sort of dealing with the enterprise system and all of the sort of Christmas tree ornaments that hang off that to make it work and dealing with all of the providers. When you come to that group of people and say, here are a hundred new apps for very specific purposes. This one helps you manage patients who've had a hip replacement better, and this one helps you manage people with diabetes better, and this one is, you can imagine. What that enterprise group of people says is we don't have time, we don't have bandwidth, and our world is our EHR.
And that new app you're thinking about, I think the EHR is gonna develop that thing in the next three years. We're gonna wait for it, 'cause that is their worldview. So you need actually a whole different arm of your enterprise system to be the group that looks at the external environment, looks at what Google, Apple, Amazon, you name it, and all of the startups are doing, vets it, figures out in the system, what are the dominant naughty business problems that cannot be solved by the EHR? And now, these days on top of it, what is this whole world of virtual care look like? And telemedicine is just a way of replacing the 15 minute visit.
The bigger deal is all of the digital tools that are helping patients manage themselves at home. How do they interact with us? So we have established a whole separate group of people. It's called the Center for Digital Health Innovation, whose job really is to monitor that outside environment and to be the parts of the enterprise that takes a really complicated problem that is not going to have a single tool solution and has a time horizon of more like three to five years rather than next year. And at the end, they're gonna build and they're already building a set of sort of tools and competencies that is going to be stitching together here's something the app can do, here's something that's not on the market we have to build it, here's something that a cool startup is doing, and we're pretty convinced they'll still be in business in two years.
Let's do that. Here's something Google or Apple is doing. We're gonna weave it together.
If you're a big health system , you're gonna have to have that set of people incompetencies, 'cause your enterprise people just are not gonna be able to do it. And that's a new problem. - Well, Dr. Wachter, thank you so much for your thoughts and your wisdom over the years, leadership in this space. I see that we're out of time.
And so we'll go ahead and stop this right here, and thank you so much for the opportunity. - Thanks for having me. - All right, well, welcome back. We'll come back to cover some of those topics discussed by Dr. Wachter,
but now it's my pleasure to introduce our next panelist with us here today, Dr. Ashish Jha. Dr. Jha is recognized globally as an expert on pandemic preparedness and response, as well as on health policy-related research and practice. He's led groundbreaking research around the Ebola, and is now on the front lines of the COVID 19 response leading national and international analysis of key issues advising state and federal policy makers. Dr. Jha has published more than 200 original research publications and procedures journals such as the "New England Journal of Medicine", "JAMA", and the "BMJ".
He's a frequent contributor to a range of public media. He's extensively researched how to improve the quality and reduce the cost of health care focusing on the impact of public policy nationally and around the globe. Dr. Jha is currently the dean of the school of public health at Brown university. Dr. Jha, welcome. - Brian, thank you so much for having me here.
And great to follow Dr. Wachter, and that really interesting conversation. - Absolutely, so you've heard some of his thoughts just now related to the pandemic. I'm wondering from your perspective, what does the COVID pandemic teach us about what our nation needs in the digital health strategy for all of us? - Yeah. Yeah, and I did hear Dr. Wachter's comments and I've known him for a long time.
I agree pretty much with everything he had to say. I will say that when I look at the pandemic and so taking a bit of a step outside of the health system, and when I look at the pandemic and think about how the pandemic has been managed, there are few thoughts that I think are really important for all of us to be thinking about. First of all, we have a public health system, so not the healthcare system, but the public health system that tracks and manages infectious disease outbreaks that is horribly outdated certainly in terms of the way it manages information. Even today across the country, people are getting tested for COVID.
At the end of the evening or tomorrow, you'll look at the CDC data on how many infections there were today. A chunk of those come from basically a lab doing a test, printing out a result, someone faxing that result to a state department of health where somebody keys in the result by hand. That's in 2021. That is not a 21st century public health information system. So one major part of the digital kind of revolution is that the public health system has largely been left aside. And that needs to change in a very, very dramatic way.
That's one of the reasons we were not able to create a more effective response last spring and summer. Let me make one other kind of big picture point and I'm happy to dig into details. When I think about the pandemic and what I track on almost a daily basis, yes, of course, I look at CDC data and state department of health data on cases, and tests, and hospitalizations, and deaths, but I also attract Google mobility data which Google makes widely available so you can look at how are people actually behaving. I look at OpenTable reservations data. OpenTable has started making their restaurant reservations not cause I'm trying to figure out what restaurant can I get into, but actually what you have seen throughout the entire pandemic is as infection numbers start rising people start changing their behavior and they're less likely to set up basically go to a restaurant.
And so you can see OpenTable reservation starts falling well before public health measures come and kick into place. The point of that is that there is all of this data out there to harness for public health monitoring and response. And it's not being done by our public health agencies. CDC isn't doing it, state health departments aren't doing it. And so we have got to come up with a strategy that brings all of this together to make us far more effective than we've been. - I wanna dig into one of the things you just said there.
So you mentioned, the quintessential example of how our healthcare system is broken with related to faxes. How is that possible? Why is that the state of the art in public health lab reported today? - Yeah, and I will say, there are a lot of state public health labs that are doing a great job. And again, this is not meant to be a critique of all of them, but there are still a lot of places that have not built the connectivity between laboratories and state health departments. The connectivity between state health departments and healthcare delivery systems is really quite weak obvious. So I have been calling for 2021 being the year that we finally retire the fax machine in the healthcare system. I think we have the ability to do it.
It's certainly a little bit about under investments and lack of funding, but I don't think it's just that. There are broad problems with technology adoption in the public sector that are complicated, but we really do have to move on from the current approach. And we've got to really modernize our public health delivery system, public health system. I will say we've made a lot of progress on the healthcare side of things.
So the healthcare, the HITECH Act, the health IT for Economic and Clinical Health Act passed back in 2009, has done a lot to bring IT systems in health care up to a different level. Now, we can talk about it hasn't paid off in all the ways we had envisioned, but clearly has digitized the healthcare delivery system. That same set of investments and the same set of metrics and forces have not been applied to the public health system and it shows in terms of what we have. - Got it. I wanna come back to the non traditional data streams here in a moment, but for right now, I wanna turn to our second panelist. I'd like to introduce to you Dr. Nigam Shah.
Dr. Shah is professor of medicine in the biomedical informatics at Stanford University. He's the associate CIO for data science at Stanford Health Care as well. He's also a member of the biomedical informatics graduate program as well as the clinical informatics fellowship. Dr. Shah's research focuses on combining machine learning
and prior knowledge in medical ontologies to enable use cases of the learning health system. Dr. Shah has received multiple teaching awards, including the AMIA New Investigator Award, and the Stanford Bio-science Faculty Teaching Award. Dr. Shah, welcome.
So really the same question I stated out to Dr. Jha to you. Based on what you've seen happen in the COVID pandemic, what does that teach you, teach us about what we need as a nation for a digital health strategy that will serve all of us? - That's an excellent question, Brian. So I'll break it down into two broad axes.
One is the healthcare access that Bob Wachter spoke to us about, and then the second is the public health access which we just heard about. And to put it very sort of concisely, public health is about getting some data on lots of people so that you can make policy decisions. Whereas the digital health Wachter was talking about was getting lots of data on one person to make the best decision for that person. Now, the requirements for both of these goals are different. We can focus on what is common to them, and what is common to them is clarification of who is on first. So for example, in public health, one health system that I happen to be a part of, you have to comply with the guidance or requirements of six different counties who all have their own different rituals about how the electronic case report form work.
The case report form may not be electronic, should be submitted to them. And all of that has to be done practically without funding. Whereas on the healthcare side, every digital innovation we do materially improves care or reduces cost. And so in terms of like guidance that is applicable to digital health in both spheres, I think the fundamental thing we need is clarity on who is on first and who owns the data. If we can get those two things right, we can have a lot of progress in terms of actually using those data sets for a lot of different uses. - So you're both bringing up data sets, traditional data sets, non traditional data sets.
So, Dr. Jha, I'll turn to you on this one. So when we think about those non-traditional data sets and the kinds of stakeholders that would need to come together as part of public private partnerships or other kinds of coalitions, what does that look like for you for the next two to five years? What kinds of public private partnerships do we need to drive digital innovation forward in this space? - Yeah, it's a great question, Brian. And I have a couple of thoughts on this. I mean, first of all, what I have seen is a lot of private sector companies, a lot of businesses really willing to engage on this, right? I mean, Google didn't have to put out it's mobility data, OpenTable didn't have to start making this stuff. They're not making any money on it. They did it because, I mean, you can ask them ultimately their motivation, but the way I understand it and the way I've seen it interacting with them is that they wanted to make their data available, 'cause they thought it was useful data for public health and others to engage with.
So I think the first of all, I think the government has to, as it begins to do this outreach they're gonna find a lot of willing partners. We're gonna have to try to set up some rules of the road. There are some really important policy questions that are out there. Google can release a lot of data in aggregate, as Dr. Shah said, like, public health people wanna look at large amounts of data on large numbers of people.
Google can do that. Google can also give you a lot of individualized data. And so then the question is, do you want that? And what are the rules of the road of that? And how do you deal with anonymity? And do we really want Google and the government working together on all of this stuff? You can imagine how you quickly go from, I'd love to look at the level of mobility in Boston, Massachusetts to wow, this stuff is starting to make me nervous. And what I wanna do is I think we've got to engage in that conversation.
I wanna do this very openly and publicly. We should have a discussion as a country, what kinds of data do we want people to have, and what kinds of aggregation do we have? Each of these data sources individually out there and then me aggregating it in my brain is one thing, all of the data being pulled together in one place and then being that starts setting up a whole different set of challenges. So I think these partnerships are inevitable.
They have to happen. I think we're gonna be much worse off if we don't, but they raise a series of questions that really need to be addressed. - So let's stay on the topic for just a moment.
You talked about pulling large quantities of data around potentially one individual, different data sets, where they're going, where they've been, predictive analytics all around that. Lots of people have talked about the sensitive topic about patient privacy, the privacy related to one's health data, or other non-traditional kinds of data. As data becomes more digitized, right? As Dr. Wachter's vision becomes more of a reality with health data, what are some of the concerns and equities that you think we need to address in that space? - Is that for me, or for Nigam? - Sorry, that's for you, Dr. Jha.
- Yeah, yeah, yeah, yeah. Great, yeah, I'm happy to do it though I'm actually interested in Dr. Shah's opinion on this too, Brian. - Please call me, Nigam. - Nigam's views on this as well, but let me say that look, what I think we'd all be concerned about is making sure that people really understand what they're signing up for and what they're signing over.
There is a lot of work to be done on issues around medical literacy and people just like a lot of times, people don't really understand how their data is being used. And having long data use agreements where somebody just signs a five page thing, but they've never read it is not consent. So part of the reason for having these conversations out in the open is that we need to as a country to have some conversations about what kinds of what ways are we gonna be able to use these data. But to your issue around equity, look, this stuff often begins with like the young, healthy people who wanna like completely digitize their lives. That's fine, but pretty quickly we're gonna get into an area where if there are really powerful tools that are to be had and I believe they are, you know that wealthier people, people who are more educated, people who go to healthcare delivery systems that are more financially successful are gonna reap the benefits of this stuff much, much faster. So I think we have to understand that that is what we're setting ourselves up for, and then figure out how do we mitigate, or ultimately, ideally avoid that.
And that's gonna require a proactive policy action. It's not gonna happen naturally on its own. - So, Nigam, what are your thoughts? What do you think about that? - 100%. I think we can't afford to let the industry drive regulation.
The government and the people have to choose for themselves first. And so continuing with that sort of two axes that I set up, digital health for public, or digital technologies for public health, and digital for health care, right now, we have a decent amount of regulation though outdated for the healthcare side. In fact, it's outdated enough that under the guise of HIPAA, we have created a massive and large market for buying and selling healthcare data, which most people don't even know it exists. We don't even have that on the public health side. So in some sense, that is behind. And at the same time, we have this notion that digital health or digital tools plus large data sets will give us AI or algorithms that guide decisions.
But again, on the both axes, public health and on healthcare, AI works when there's a clear feedback loop. So the example Bob Wachter gave about a glucometer works beautifully because the response is injecting insulin or titrating your drug and you see the effect immediately. What is the analogous closed loop feedback for public health? We don't have that.
And so the regulation and the contract that would need to be in place between the person that is generating the data, or as the object of study, either some data sets about me and my restaurant visits or my entire medical record in the healthcare setting, I am the focal point as an individual. What is the contract between me and society for using my data for either public health or for healthcare? And I'm one of the sort of the tailenders I say, if I want to benefit from other people's data, it is my duty to contribute my own. I mean, that is the fundamental notion of a commons and we can get away from that. I can't benefit from other people's data and choose not to share my own.
I mean, it just doesn't compute for me. - Dr. Jha, I'd love your response and thoughts on that. What does the social contract for individuals as it relates to public health data look like for you? - Yeah, I wanted to jump in.
I think what Nigam said was exactly right in that we sort of have this mental model of every single individual needs to decide. And on some level we do believe in autonomy and that's fine, but it doesn't really work in this context, right? And we've seen this. Like, if you look at the abuses, for instance, from Facebook, their argument is we have end user licensing agreements and, therefore, we can do whatever we want except most people had no idea what was actually being done.
And so I think kind of this idea of everybody kind of signs. You have a choice. You either use Google and sign away your life, or you don't use it at all is a false choice and it's completely acceptable, unacceptable, not acceptable, unacceptable in a democratic society. So what you need is much more of a public discussion about how do we wanna use these kinds of data and what do we wanna allow platforms to be able to do. And that's a place where really ultimately has to be our policymakers, that has to be public conversations.
And I understand that some people will feel more comfortable or less comfortable with where we land, but if we're gonna do this as a collective, we have to have a very different set of conversations then each individual person's signing these licensing agreements. - I wanted to... Go ahead, Nigam. - I mean, there are precedents. So the economists routinely study data on labor markets, on our tax filings, and so on, and it has helped us create a more inclusive financial society. Why can't we not do that for both sides of healthcare, the public health and the individual? - It's a great question.
I wanna turn to another really important area of equity as it relates to individuals that, Dr. Jha you mentioned the quintessential well-educated young, healthy, fully digitally connected individual millennial and the individual that is from a vulnerable underserved community that might not be as well connected digitally. What sort of priorities, what sort of strategies Dr. Jha would you recommend and think that we should have as part of the strategy in digital health to address the concerns that, well, great, if you get into digital health, you're just gonna increase the digital divide and it's just gonna be worse off for those that aren't connected digitally? - Yeah, that's a great question.
I have a couple of thoughts on this. I mean, one is to ask why do these digital divides get created? And one part of it is technology is often developed by people who are young, and healthy, and so they build it for people like themselves. And that means that if you want more inclusive use of technology, you have to bring different people into the conversation, into the development of that technology.
It can't just happen because we say to Silicon Valley, Hey, make more inclusive technology. That's a nice idea. We can say it. I'm not opposed to saying it. It just, it's not gonna work. It's not gonna be enough.
Second is all of the evidence I have seen suggest that communities of color, poor communities have a huge hunger for technology and the use of technology. In fact, I believe that digital technology should be the great equalizer in actually helping our society get much more equity, because it makes a lot of things higher quality, it makes things lower costs, it makes things more accessible, and, therefore, it should be a major force for good, but it doesn't happen naturally. And so part of it is I think the responsibility of developers to be thinking about what can we do to create a more inclusive development kind of ecosystem. And then part of it is policymakers helping spur on adoption, providing financial incentives for things like broadband and other things that are kind of fundamental to technology adoption, because without broadband access, for instance, you're gonna have all the apps you want, but if you don't have good access to the internet most of them aren't gonna work very well. So there is a role for policy makers and there's a role for the private sector. - So I guess to dive into that a little bit more, what would be some of the specific steps that you would recommend big tech take? Is it hiring more staff from underserved communities? Is it building offices and facilities in rural communities to employ those people or speak to them directly? What does that look like exactly? - Yeah, I mean, certainly those things.
I think it is also really engaging with those communities to understand what the needs are, but at the end of the day it's not just about focus groups. It really is about having people in the companies who have connections to those communities, have lived experiences really understanding. One of the things that we've all we have learned about technology over the last 10, 15 years, is that technology takes off and is useful when it works with the way we live, right? And when it works in our lives, it's not just some cool, funky thing that we all like, but we don't ever use.
Well, you have to understand people's lived experience. You have to understand what goes on in a daily life in somebody. And to understand that, yes, it's about having more people from those communities. It's about engaging those communities much more deeply. - Gotcha. And so, Nigam, I wanna put a little bit of a different twist on this.
So you're an expert in artificial intelligence and machine learning techniques. Kevin Scott, the chief technology officer of Microsoft wrote a book about how we as a society need to have a strategy to ensure that artificial intelligence serves just as Dr. Jha was saying, not just those people in Silicon Valley, but those in rural America as well. What are your thoughts? How can we as a nation rally around a strategy that does that? What would we have to do? - So I think about that a lot. And one of the conclusions I'm arriving at, and it's not sort of set in stone conclusion this is still a process, is I look to like the FAA, or the National Highway Safety Transportation Board, transportation safety board, I'm sorry, there are standard tests that we put that equipment through, planes, horses, cars that are designed to ensure the safety of the users. We have nothing like that for apps, absolutely nothing, or for algorithms.
As you know that today, there was a headline there's 350,000 digital health apps, 90,000 of them just in the past year. - Wow. - And self-regulation, I don't think is gonna work here.
And so what we need to do is, again, coming back to policy and legislation come up with the desiderata, that is at the minimum an app that is designed for XYZ needs to do these three things and then we have a test for that. So minimum competency test for the fairness, the usefulness, and reliability of algorithms that claim to improve health or healthcare. - Gotcha.
So one last question, Dr. Jha. So we've talked a lot about policy, we've talked a lot about public health and empowering individuals with digital health tools. What sort of calls to action would you challenge our policymakers, our elected officials around digital health to make public health more empowered with the kinds of digital health tools that are on the traditional healthcare side? What would be the call to action you would issue to them? - That's a great question, Brian. A couple of thoughts went through my head as you were speaking. One is if you think about this pandemic, it has just been devastating for our country, for the world. And as we begin to think about how we identify or prevent the next pandemic, as we begin to think about how do we recover from this, I think one of the things that we have learned is that technology can be an incredible aid.
It can teach us things, it can give us early insights, and yet most of our government isn't geared to use that technology. Most of our government doesn't know how to access it. And so what that means is over the pandemic, I've worked with a lot of private sector companies where I would look at the models that they were building for where case numbers were going, hospitalizations, and deaths, they were almost all consistently much, much better than what the CDC was using. That's unacceptable, right? And that's because the CDC is very limited in what it can do, and how it works, and how it works with private sector.
And I think the rules have to change. I think the government has to become much more open to using private sector data, has to understand the limitations of it, of course, there has to be some amount of rules around conflict, but the government cannot be the sort of poor person out who's can't access what much of the world is being able to access. And so what I would say for policymakers is lets envision a very different public health agency, a public health infrastructure and network that is deeply engaged with the private sector, pulling this kind of information, using it certainly in the next pandemic or preventing the next pandemic, but between now and then we have plenty of health problems to deal with.
And so being able to engage with the private sector to solve those public health problems would be a fabulous way to get ready for the next pandemic. - Agreed. Well, thank you for that, Dr. Jha. And so I wonder if we have time for maybe just one or two questions, and I wanna turn to Dr. Shah Nigam.
So there's a question coming in from an audience member. How do we strike the right balance between personal health data protections and access to quality health data for the research community for building AI algorithms to serve all of us? - That's an excellent question. Something we struggle with every day. And I mean, my take on this, and this is a personal take, I think we got to get over the fact that de-identification as the word is commonly understood is a myth. We have to assume that we cannot guarantee anonymization and design permitted use cases under that assumption.
- Do you think it's possible to have de-identified data used in a data set for ML purposes and for that data to be not guaranteed, but fairly certain that it can't be re-identified? - So it depends on what definition of de-identified. If we use HIPAA's definition as of 1996, possibly under the safe harbor approach of de-identification. But I mean, there was a recent study which showed that just looking at CT scan and MRI cross sections, you can reconstruct somebody's face. Yeah, so the thing is the general interpretation of the word de-identification, and what is possible in some MIT, Harvard, Stanford lab is always going to be in congruent.
And we have to be cognizant of that, that there will always be computational approaches that can circumvent today's definition of de-identification in a way that you meet the definition, but you can still identify who the person is. - Yeah, yeah. - And so we have to design use cases assuming this is possible. - It's a great point. Well, I wanna be cognizant that we're reaching at the top of the hour and that we're pretty much done here with this power hour.
I appreciate everyone joining us for the call to action for a national digital health strategy. Thank you for joining us today, and thank you for our panelists, and esteemed guests. We hope that you'll take some time to download and read the national strategy for digital health. The links to those papers can be found at the top of the chat box in your window. We hope you'll reach out to the digital health team here at MITRE as well with feedback, or to schedule a discussion for further engagement in conversation around the...