The Human Diagnosis Project, also known as Human Dx, is one of the eight competitors up for a $100 million MacArthur Foundation grant aimed at achieving a feasible, sustainable solution to a pressing global problem. In this Business of Giving interview, Jay Komarneni, the organization’s founder and chair, outlines its plan to build out a digital system to improves access to high-level of care for hundreds of millions of medically underserved people.
Mr. Komarneni likens the project to Wikipedia, or the open-source operating system Linux. Like those ventures, Human Dx aims to harness the power of technology and the collective expertise and skills of an open, global community. In this case, the community is medical professionals, and the goal is "to answer the essential question of human health and well-being, which is: When you or someone you love isn’t well, what should be done?"
The roots of the project are personal for Mr. Komarneni, who was born with a congenital heart defect. Growing up "in a family of physicians in one of the richest countries on earth," he benefited from the insight of specialists who eventually corrected the defect with open-heart surgery. With Human Dx, a safety-net clinic or primary-care physician faced with a challenging case can go far beyond a referral or "curbside consultation" with a few colleagues, tapping into an accessible, ever-expanding pool of knowledge for every step of care and circumventing delays that could be life-threatening or unnecessary treatments that can bankrupt a family.
The system already has contributors from more than 60 countries covering 40-plus specialties, and applications that range beyond specialized diagnosis to wellness and prevention, routine care, even fraud detection. MacArthur’s $100 million would be a big step toward a world in which the people with the greatest health-care needs can get the right help, sooner.
Denver: And this evening’s semi-finalist is the Human Diagnosis Project, also referred to as Human Dx. And here to tell us about it is their President and CEO, Jay Komarneni. Good evening, Jay, and welcome to The Business of Giving!
Jay: Denver, thanks so much for having me!
Denver: Congratulations on being named as one of the semi-finalists of the 100&Change competition. Give us an overview of the Human Diagnosis Project and what you hope to achieve.
Jay: Absolutely! Thanks. Denver, I think what the Human Diagnosis Project exists to do is to answer the essential question of human health and well-being which is: When you or someone you love isn’t well, what should be done? This is a question that every single person on the planet struggles with many times during their lifetime, and our goal is really to help answer this question for all and forever.
Denver: What was the impetus for you to start this, Jay? And were there any platforms that inspired your model?
Jay: The story of the Human Diagnosis Project actually starts with the day I was born. I actually was born with a congenital heart defect and was able to get access to the best care and the best specialists when I was a teenager and had to have my heart defect corrected with open heart surgery. If I didn’t grow up in a family of physicians in one of the richest countries on earth, I wouldn’t have had access to that insight. We really believe as a team that everyone in the world should have access to the world’s collective medical insight in order to get better answers to those questions.
Denver: And this is not really just a “safety net” for those people who need to go to an emergency room. A lot of this is focused around specialty care. Would that be correct?
Jay: The proposal that we had put together for MacArthur in conjunction with the American Medical Association, the American College of Physicians, the American Board of Medical Specialties, and the American Board of Internal Medicine is specifically to use the Human Diagnosis Project to improve specialty care for the nation’s underserved. That being said, the system that we’re building ultimately can help every single person on the planet with both primary care and specialty care. As you may know, a billion people on earth lack access to even basic health care, and a hundred million people are put into poverty as a function of their health care cost. So this is a much bigger problem than just the problem we seek to serve here in the US, but we think that this is a tremendous opportunity to help begin building the system, and using it to help the people who need it the most.
Denver: Let me see how this might work. Let’s say I’m an attending physician, and I come across a challenging case, and I’m not exactly sure what it is or what I’m looking at, but I’m a bit concerned. What would I do?
Jay: The way that this works typically is one of three things happens when you’re a primary care physician and you’re trying to get a better answer to your case: (1) you actually do what’s called a curbside consult, so you ask other physicians what they think– who you know and are done in person; (2) is you do something called an electronic consult where you actually ask someone through your existing electronic health record or system; or (3) you do a referral. So the issue becomes that when you’re uninsured, you’re really making a choice between two tough places; you’re deciding whether or not to delay necessary care… and potentially get sicker, or potentially pay for care that may not be needed and go into poverty as a function of your costs. There are 10 million people in this country who are in poverty because of their medical costs.
So, imagine that you’re making that decision. Well, as a primary care physician who’s helping people in the Safety Net, 90% of those Safety Net centers cannot get access to specialists. If you can actually provide them insight through a system like Human Dx, you can actually ensure that only the people who really need care are the ones that are getting care. So that when they’re paying for it, they really need it. And then you’re actually freeing up specialty capacity to help the patients who really need help. So the opportunity here– and the way that Human Dx works to solve this problem– is when a primary care physician goes to the system, they basically can encode and organize the major details of the case, post it to the system, and then have other specialists pontificate on that case. Then they can get insight much faster than they otherwise would’ve been able to by doing a traditional referral or e-consult.
Denver: How many of these cases can be addressed through electronic consults?
Jay: Well, I think what’s exciting is that the literature shows anywhere from 30% to 50%.
Denver: That’s a lot.
Jay: Exactly! And here’s what’s exciting about that. A lot of that research is 15 years old. How much has the internet and technology changed in the last 15 years? We actually believe it could be much higher with real-time messaging, video chat, images, audio – all of these different tools can actually make that fraction of visits even higher. And then what’s exciting is not only does it help on that fraction of visits, the 30% to 50% can be completely eliminated with respect to not requiring someone who can’t afford the care to get it. That capacity can then just be reallocated towards people who actually need that care. So, we can close the specialty care gap not just in the 30% to 50% range, but potentially much more.
Denver: Let me ask you about the internet. Boy, there’s an abundance of medical information on it. I would venture to say that everyone in our listening audience has gone online at one time or another to check out a medical condition. So, how will this project differ from existing resources and solutions that are currently available?
Jay: There’s a bunch of issues with existing sources of information, and the reality is that whether or not physicians like it, whether or not patients like it, everyone is on the internet self-diagnosing. It’s the third most popular activity after search and email. Yes, if you go to the few internet studies, which is kind of fascinating. Now, interestingly, they’re getting information that’s not the best to make these decisions.
To give you a sense, we recently did a study with Harvard Medical School. We found that individual physicians still dramatically outperform machine-based systems that are available to everyday people. Part of the reason that is is because the data that’s used to create these systems is very flawed. So, one is: it’s highly unstructured. It lacks clinical nuance. So you might be able to put in your symptoms, but you can’t put in the intensity, frequency, duration, location, aggravating factors, mitigating factors of those conditions. You can’t put in lots of other salient details like your physical exam results, your lab tests, your social history, where you grew up, your medical history – all of those other kinds of components which are tremendously important at coming to a good answer.
And so what’s different about the system that we’re building is not only is it a platform technology – so one application is a clinical decision support system; one application is a symptom checker; one application is detecting fraud… which creates huge costs to the entire system in terms of paying for care. Twenty percent of potential Medicare claims may be fraudulent. We actually believe there’s a fundamental system and a fundamental data structure that we’re building through Human Dx that can ultimately help anyone, anywhere, at a significantly reduced cost.
The word diagnosis in the Greek root means ‘to know apart,’ and ultimately, we thought that was a very fitting name for a project which is ultimately about any clinical decision.
Denver: Let me ask you a little bit about your name. The project suggests your focus is diagnosis and not for instance, treatment. Why that distinction?
Jay: It’s a great question. The word diagnosis in the Greek root means ‘to know apart,’ and ultimately, we thought that was a very fitting name for a project which is ultimately about any clinical decision. We’re building a comprehensive system, created with and led by the global medical community and inspired by projects like Wikipedia and Linux, that ultimately maps the best steps to help any patients. So that’s really: problem to triage, to diagnosis, to treatment, to management, to discharge, to on-going care, wellness and prevention.
Now, for better or worse, the human clinical decision project or the human decision project didn’t quite sound as good, so we decided to pick a word that we felt represented the entire continuum of care but did so in a way that was a little bit more appealing.
Denver: You mentioned before that the primary care provider takes what they see and puts this case up on your site, and specialists from around the country and around the world contribute to help with the answer. Why would the medical community contribute to this project? What would inspire them to do so?
Jay: It’s a great question. I think almost every doctor ultimately started their journey in medicine with the desire to help others, and as you know, everything in medicine from the Hippocratic Oath on down is about serving patients and helping them. I think many physicians are also frustrated by the lack of care that so many people in the world have – 50 million plus in the US, perhaps a billion plus worldwide – and that doesn’t even include the people who don’t have high-quality care.
So when we look at this, we realized that physicians have a deep interest in helping close this gap and making care better for all involved. And so that’s kind of Part one. I think Part two is: this is actually a very powerful resource for clinicians themselves – for training, for teaching, for learning themselves. There’s almost never been a system which could provide medical education to any person on earth at essentially zero marginal cost.
Denver: That’s pretty sweet. I know you’re in your very early stages, but what does your participating universe look like at present – the number of contributors and the number of specialties in countries that are represented?
Jay: We now have thousands of contributors from 60-plus countries and 40-plus specialties. I think what’s really exciting about that, too, is the people who are contributing are doing extremely complex, hard cognitive work to contribute to the system. They’re not just posting on Facebook or upvoting or liking something. They’re really doing complex cognitive work on behalf of the project, very similar to how software engineers built Linux. And over a period of about 25 years, only about 15,000 software engineers have built Linux, which is now the world’s most important open-source software project and has created in a crazy way over $100 billion of economic value just through the efforts of people who were contributing their time voluntarily.
Now, to give you a sense of scale, there are about 12 million software engineers in the world, and there are about 15,000 people who’ve contributed to Linux over 25 years. There are about 12 million doctors in the world, and over 5,000… close to 6,000 have already contributed to Human Dx in two years of being an open project. And so we think there’s a bright future ahead for where we’re going.
Denver: Very impressive! Well, a project of this magnitude and size is going to have to evolve over time and into different phases. So, Jay, what do you see as some of the near-term benefits? And what do you envision the long-term potential to be as a result of this initiative?
Jay: I think one of the first major applications of Human Dx is in enabling and empowering clinicians to make better decisions at the point of care, to essentially reduce medical errors, reduce unnecessary task procedures, services, etc., and then also recommend when a certain type of visit should happen or not. Now, why is this so crucial to our work with MacArthur? Well, as I pointed out, people who seek specialist care but are uninsured face a very tough choice. They have to either delay necessary care and get sicker, or potentially incur the cost of care that may not be necessary. So our view is: “Let’s focus on the 50 million here in the US and really improve care for those people as soon as possible. And then let’s then scale that to actually serve people all over the world – first, the billion people-plus in the world who are underserved… but ultimately, anyone on the planet.”
And that actually is a nice segue into why MacArthur is so crucial here. If we had the opportunity to use funding– like the funding from MacArthur 100&Change– to actually pursue this project, we could ensure that this care first goes to the people who need it the most. Otherwise, we kind of have to do the classic technology meme, which is start serving people who are better off, and then hopefully, over time, serve people who are worst off. The sad part about this is there are one billion people who still lack access to health care. There are one billion people who don’t have clean drinking water. There are a billion people who don’t have reliable food access. So people say this kind of idea of: let’s provide care and then ultimately, hopefully it will trickle down to others – the reality is that this rarely works. You have to build solutions from Day one which have the ability to impact even the worst-off.
Denver: And MacArthur would let you turn this whole thing upside down, which is the way you would like to do it.
Jay: Absolutely! I think that’s what’s so exciting: the opportunity to serve the underserved first, which is rarely what happens with transformative technologies. That’s something that could be really exciting here.
Denver: You know, I’m a little curious about these clinical case contributions that are going to be used to build this project. How are you going to validate their quality and their accuracy?
Jay: I think what’s really interesting is the way we’ve built the entire system is to create a distributed, anonymized verification approach. The way it essentially works is when an attending clinician posts a case to the system to get input on it from other physicians, multiple other physicians attempt to independently solve that case. And when I say solve that case, they don’t just provide their answer, they actually show their work. So as they’re going through the case, they’re revealing information; they’re updating their list of possible answers. Then they’re revealing more information; then they’re updating their list of possible answers again… on and on until the end of the case. And they’re doing this in a way where they don’t know who else is solving the case. They may not even know who contributed the case.
So what we’re, in a funny and interesting way, doing is getting a distributed, anonymized verification to see how different people come to similar conclusions with similar data. And so at the very least, we can see that multiple stakeholders have then looked at this case similarly. Then by actually ingesting external data sets, such as clinical guidelines, research data, et cetera, we can actually see: is the answer that the collective is arriving at.. the one that is actually the correct answer based on additional information as well?
Denver: And as you said, if you can’t even see who made the contribution to that answer, I guess that addresses my next question– that being of privacy, both patient privacy and contributor privacy.
Jay: Yes. So clinicians in the system can both post and/or answer cases as anonymously as they want to, but then the key is that patient data in the system does not include personally identifiable information. So it doesn’t include names or faces or social security numbers or anything else that could be personally identifiable to date. That being said, as we begin implementing in the Safety Net, we will also be handling that type of information. But that information will not be passed necessarily to people who are pontificating on the case unless it’s specifically important and unless it’s done with the permission or provisioning of the patient or their clinician.
Denver: You know, I mentioned in the opening that the Human Diagnosis Project is also referred to as Human Dx. What does the name “Human Dx” represent?
Jay: Yes. So the reason we also have the short form of the name is because there’s actually a deep connection between medicine and computer science and math, which just all come together to build the Human Diagnosis Project. So Human Dx is the kind of literal name. Dx is the shorthand for “diagnosis” in the same way that Rx is the shorthand for “pharmacy” or Tx is for “treatment.” It also is this analogy or interesting metaphor to the summative potential of many human beings.
If you remember from calculus class, the integral is typically ended with a Dx, so the idea of multiple humans working together, the summation or the area under the curve to actually create tremendous potential. And then lastly, there’s a nice intersection of how the two tie together because the letters “Dx” actually also in math represent “differential of x,” and differential is actually the term used by physicians to account for their list of possible answers when they’re thinking about how to pontificate on a clinical case. So there’s a connection to both the math side differential of x, but also the clinical side differential of x.
Denver: Let me ask you a bit about: if you should be so fortunate as to be awarded that $100 million, how would you go about your work differently than you would otherwise? You’ve already talked about inverting the pyramid some, but tell us: what you would do with that $100 million, and how that would accelerate this program.
Jay: So, I think right now, we have a decision to make. We have to either focus on becoming self-sustainable from a commercial standpoint so that we can ultimately grow this project, or we can focus on really creating impact for the people who need it most. I think there’s kind of that bifurcation that’s simply a matter of pragmatism – if we want to stay alive, we have to be commercially self-sustainable. So the difference here is instead of focusing on integrated managed care systems, or accountable care organizations, or insurance companies who may be able to benefit from the services we offer, we can instead focus on infusing that money directly into the Safety Net to have them implement the solution, and ultimately use this to help the nation’s underserved.
Now, I think why that’s exciting and interesting is a large portion of the grant that we would potentially receive would actually go directly to funding the US Safety Net. It actually wouldn’t in any way fund even the development of our technology; it would really go primarily towards systems and capacity building and implementation. And I think that’s also kind of a unique feature of being a purely software and data project– is that the actual overhead to run the project ends up being quite low, relative to the value or resources that we could put towards implementation.
Denver: Got you! Aside from the financial resources, which are going to be needed to do this in the way that you would like to do it, anytime you put something together like this, which is so ambitious, there are certain aspects of it where you have greater concern than others. What would be your biggest challenge with this project that you see, other than the money?
Jay: That’s a great question. I think when we think about the biggest challenges to growing the project, I think the one that we constantly come back to is: How do we continue to make this such a powerful resource for clinicians that they use this constantly in their everyday work?
So right now, Human Dx is a tremendous way of contributing your knowledge to a social mission. It’s also a very great way to train and teach physicians. We are now building the system such that it can be used for specialist insight and ultimately to get help from other clinicians. But ultimately, our vision is: How does this really incorporate into every facet of the clinical workflow? And how do we make this a system where every single time you’re looking for information, or you’re trying to collaborate with other clinicians or potentially even patients, you use this system and not another system? And so I think getting down that product road map and really making this so compelling and powerful that clinicians want to use this– not just to help out, and not just to learn– but also to use it for every single thing that they do. That’s the thing that we think about constantly in terms of: How do we make that possible?
Denver: Jay, this system has tremendous potential to improve the medical enterprise here, both in this country and around the world, but what does it mean to an individual person? How will it change someone in our listening audience right now? How can this potentially change their lives and that of their families?
Jay: I think the way to start answering that question is to actually talk about the stories of the team and how we came together. What’s interesting is almost everyone on the team has a friend, a relative, a family member, themselves who’s had a major clinical problem. One of our team members has a daughter who was born with a kidney problem. Another one has a mother who’s suffering from rheumatoid arthritis. I mentioned myself. I have a congenital heart defect that required open heart surgery.
And then everyone also can look to their neighbors and their friends and see the same thing. You find someone who doesn’t have health insurance to cover their cancer treatment, an immigrant who hasn’t seen a doctor in more than 10 years because they can’t afford to, out-of- work mothers who can’t get medicine for their children – this is all a function of the fact that healthcare is too inaccessible and too costly. There is a huge percentage of people who aren’t able to get the care that they need and deserve. So, if you can build a system which can supplement the great work of all the physicians worldwide and actually extend their capacity to be able to help more patients, all of those instances that I just described could be significantly better dealt with.
Denver: And it really does have that exponential potential because it takes a little bit of time to build up, but then it just grows and grows and grows at a geometric rate.
Jay: That’s exactly right. And the system continues to get better and smarter for all involved. And I think the key is that as the system gets better and smarter, it makes everyone’s life easier– not just patients, but also clinicians because clinicians can focus more on the complex situations that their attention is better suited to. And then the simpler situations can be resolved with lower cost.
Denver: Let me close with this, Jay. You clearly have some really wonderful and worthy competition in the other semi-finalists, what case would you make as to why this project is the one that will have the greatest impact and benefit to humankind?
Jay: I think to start off, there are so many interesting and worthy projects here. I think in addition to the seven competitors, you may have seen that MacArthur also released the top 200 proposals, and all of them look astounding. And so there’s no question that there is a lot of good to be done in the world, and there are far more great opportunities than there are funds to support those opportunities for better or worse.
I think what’s different and unique about what we’re doing are a couple of things. I think number one, what we’re building is something that can literally impact every single person on the planet, and I think there are kind of some unique features of what we’re doing that make that the case. So our purpose as an institution is to elevate well-being for all. There is a subset of things that can truly elevate well-being so markedly, or notably improve well-being, and there’s a subset of things that can impact all people. But there are very few that can do both. I think what’s unique and interesting about what we’re trying to accomplish is we’re answering a question that every single person on this planet has many times during their life: When you or someone you love isn’t well, what should be done? I think when we start with helping the people who are struggling the most in the US and then abroad, ultimately, that’s kind of the initial focus. But the ultimate focus is every single person on the planet, and I think that’s a very different aspect of what we’re doing. That’s kind of number one.
I think number two is because of the fact that this is software and data, it really can scale at zero marginal cost. So this isn’t just wishful thinking; this is actually something that can get done. And then I think number three is because of the unique features of what we’re building, this can be truly self-sustainable from a commercial standpoint. And in order to create a solution that answers the question that will be relevant as long as humankind exists, you need to build a durable solution that’s not just something that can last for decades, but something that can be truly eternal. And if we can build a system that is commercially self-sustainable from the efforts of our work, I think we can not only impact every person on earth and do it quickly at zero marginal cost, we can do it forever.
Denver: Well, Jay Komarneni, the President and CEO of the Human Diagnosis Project, I want to thank you so much for being with us this evening. If people want to learn more about the Human Diagnosis Project or become involved somehow, where can they go to get that information?
Jay: Our url is humandx.org, and you can learn more about us there. And we have links to several materials. You can also just contact us directly.
Denver: Well, best of luck to you and your colleagues, Jay, in the MacArthur Foundation’s 100&Change competition. It was a real pleasure to have you on the program.
Jay: Great! Thanks so much.