The Patient We Never Designed For (Live from Cornell Ithaca, Extended Cut)

Recorded live at Cornell Ithaca. This is the extended cut of Dr. Sarah Matt's keynote on the central design flaw in modern healthcare: a system optimized for an "average" patient who does not exist.

When you build clinical workflows, EHR defaults, drug dosing tables, and AI training sets around a statistical mean, every patient outside that mean pays the cost in workarounds, missed diagnoses, and unsafe care. And most patients are outside the mean.

In this talk, Dr. Matt walks through:

- Where the "average patient" assumption shows up in clinical AI, and why it compounds bias instead of correcting it

- The design debt baked into EHRs, order sets, and risk calculators

- What health systems and physician leaders can do now to design for variance instead of pretending it away

- Why the patient we never designed for is the patient most likely to be harmed by the tools we are deploying next

Recorded live at Cornell. Audience Q&A included.

—

Resources & Links:

Get the book: "The Borderless Healthcare Revolution" is available now on Amazon and major retailers.

Work with Dr. Matt: Looking for a keynote speaker or strategic advisor? Visit drsarahmatt.com

Connect on social:

LinkedIn: https://www.linkedin.com/in/sarahmattmd/

YouTube: https://www.youtube.com/@DrSarahMatt-ClinicalRealist

Subscribe to The Briefing: drsarahmatt.com/newsletter-signup

—

Disclaimer: The views expressed on this podcast are those of Dr. Sarah Matt and her guests. They do not necessarily reflect the official policy or position of any affiliated institutions. This content is for informational and educational purposes only and does not constitute medical advice or a professional consulting relationship.

—

Resources & Links:

📖 Get the Book: "The Borderless Healthcare Revolution" is available now on Amazon and major retailers.

💼 Work with Dr. Matt:
Looking for a keynote speaker or strategic advisor?
Visit: drsarahmatt.com

🔗 Connect on Social:
LinkedIn: https://www.linkedin.com/in/sarahmattmd/
YouTube: https://www.youtube.com/@DrSarahMatt-ClinicalRealist

📧 Subscribe to The Briefing: drsarahmatt.com/newsletter-signup
 
—
Disclaimer:
The views expressed on this podcast are those of Dr. Sarah Matt and her guests. They do not necessarily reflect the official policy or position of any affiliated institutions. This content is for informational and educational purposes only and does not constitute medical advice or a professional consulting relationship.

Transcript

SPEAKER_01 0:03

All right, so happy uh Wednesday, no happy Tuesday, happy almost Wednesday. Really nice to see uh many of you that I don't get to see all that often. Uh thanks for being here today, and I'm really excited to introduce Sarah Matt, MD, MBA, medical doctor, assistant professor, health technology strategist, growth catalyst, and an author. So, Dr. Matt is a surgeon turned health technology strategist. Her work focuses on how digital tools from remote surgery to telemedicine to AI can expand access to healthcare and eliminate the traditional boundaries that separate patients from care. Through her various leadership roles across first Oracle, then NextGen, then several startups, Dr. Matt has worked around the world to explore challenges and identify solutions across different healthcare, cultural, and economic environments. Dr. Matt has seen what works, what doesn't work, and most importantly, what's possible when healthcare access and delivery is reimagined to overcome geographic constraints. Last year, Dr. Matt sat down and found a way to share these experiences with the world, writing a book, The Borderless Healthcare Revolution: The Definitive Guide to Breaking Geographic Barriers Through Technology. I'm really looking forward to hearing from Sarah over the next hour, and I hope you really enjoy this. Thank you for being here.

SPEAKER_00 1:29

Of course. All right, we got to get the dramatic mood lighting first, right? Okay, awesome. So thank you, Jen. I appreciate that. So for those of you who are here because you had no other reason except someone told you to be here, I appreciate that. For those of you who have maybe looked at my book, I also appreciate that too. But today I'm gonna start with something completely different. I want to start with a story about someone you've never met. And when we think about the person that maybe isn't someone you interact with every day, what would they look like? She's 40, she works over 40 hours a week, she's a home health aide, she has three kids, she's on Medicaid, so on paper, she's insured. She takes two buses to get to work every day. And she has a smartphone, but her data plan runs out every month in the third week. So basically, she's working on Wi-Fi when she's lucky. So in early 2001, she heard that she could get a vaccine for COVID. And she didn't know much about it, she didn't know what number to call, but she found a website that listed appointments at a clinic across town. And she called a number on the site, and she waited on hold for 40 minutes. And then she had to hang up. She called again to another one, and they said that all the appointments she saw online were in a different county. And then she tried one more time and it got disconnected. So she tried multiple times and eventually she gave up. Now, around that same time, I was doing the same thing. I'm a physician, I have a medical degree, I got way too many letters after my name. I have 20 years navigating the healthcare system. I know all the numbers. I can call my friends that are actually doctors. I know which pharmacies have things, all the cheat codes, all the billing numbers, all the formularies. I had all that access. Plus, I had a car, I had time, I could do all these things. I spent the better part of several days trying to find appointments for me and my family. And the best I could do after many calls, after getting hung up on, after getting put on hold, and then getting redirected to someone completely not the right person after being on a wait list where they never call me back was appointments 90 minutes from where I live. So I want you to sit with the distance between those two stories. I had every advantage: credentials, a car, time, expertise. And I knew how to navigate this system that's completely opaque to most people. And the best I could do was 90 minutes away. She had none of those advantages. She had a smartphone running out of data, she kept getting hung up on, and she couldn't afford to stay on. The system did not fail us both equally. It inconvenienced me, but it completely failed her. And we designed it that way. We designed that outcome into the system, not accidentally, not as an oversight, by every workflow diagram, every patient portal, every scheduling interface, every UX prototype at any tech company, we produce that. We produce things for people that look like me and not for people that look like her. So here's what I need you to understand. I started as a surgeon, then I moved into the tech sector, and ultimately I was building all the technology from patient portals to cloud for healthcare and life sciences, all the different applications. I've been building product development strategy teams all over the world. I've seen it from all sides. I also still practice medicine, so I do charity care out in a night in New York, which is a very rural place. So ultimately, I've seen the richest and the poorest. I've seen people with all the access and none of it. And here's what's happened over the last decade because it's the foundation for everything I'm going to go going to be saying after this. Between 2010 and 2023, the United States invested over $150 billion into digital health infrastructure. Electronic health records, those patient portals I terribly hate, telehealth platforms, remote patient monitoring, AI embedded X, Y, or Z, all this stuff. The explicit promise from every major industry conference, every press release, every investor deck at that time was that technology is going to close the access gap. Digital tools are going to reach the people who need it most. It didn't happen at all, actually. By virtually every aggregate measurement of healthcare access, the gap between who gets care and who doesn't is the same today, or even worse than it was back then. Not because the technology failed, because the technology actually worked pretty well, but because we deployed technology on top of a broken system and called it transformation. So for 50 years, the American healthcare system has built around a very specific patient. He's mid-30s, speaks fluent English, has high-speed internet, a data plan he doesn't worry about, great Wi-Fi, lives within 15 miles of a major hospital, has a great job where he can actually leave in the middle of the day to go to doctor's appointments. He has paid time off for those appointments, he reads at the college level, and he's comfortable navigating bureaucracy. And he trusts the healthcare system because the healthcare system has generally treated people that look like him rather well. He's in every workflow diagram, he's in every pamphlet for a new product, every consumer-facing health application, every telehealth onboarding plan. This is the same guy for over a decade. He's a patient we designed for. And he represents at most a third of the population at most who actually need care. The other two-thirds, we didn't design past them. We designed right past them, completely. And then we called the gaps in access social determinants, or we call them a health literacy problem, or a patient engagement problem, as if the failure was the failure of the patients. It's not. Here's what the actual patient population looks like. Seventy-five million Americans cannot understand and interpret a medication label or discharge instructions. 75 million. That's not a niche population. That's a design specification that we completely refuse to write. Almost 30% of Americans live in a household that doesn't speak English. Actually, this is the next stat. I apologize. 30% of rural Americans live more than 30 miles from the nearest hospital. And when you have a stroke or a cardiac event or a pregnancy complication, that determines your prognosis. And here's the one number I really want to say: 68 million people don't speak English in their home. And I actually thought that was an interesting number. I had to look it up three times because I didn't trust the citation. The percentage of healthcare encounters in those languages by native speakers is in the single digits. And the foundational assumption embedded in virtually every digital health strategy is that patients have reliable broadband and functioning devices. So they can navigate that patient portal. And it's wrong for approximately one in three Americans. One in three. Not a corner case, not an edge population, not a rounding error, one in three. We invested $150 million in tools that work reliably for two-thirds of the population, maybe. And we called that healthcare transformation. We're not failing to execute a good system. We're perfectly executing a bad design. So I've spent years studying exactly where access fails, and it fails at exactly five different points. And in my book, The Borderless Healthcare Revolution, thank you, Jen, for mentioning it, I call them the five pillars of access. The reason I name them very precisely is that the fix for one pillar is not necessarily the fix for the others. Confusing them is one of the most impossible and expensive mistakes that the industry has been making. I've watched health systems deploy telehealth platforms to solve a problem that was never about distance. It was about trust. I've watched digital literacy programs aimed at populations whose real barrier was knowledge, not bandwidth. And the wrong intervention, executed brilliantly, produces nothing. And sometimes it actually produces less than nothing because the budget that could have been used for effective designs was already spent. So the first pillar is physical. Can they get there? Think about end-stage renal disease. So this is dialysis, three sessions a week, every week, for your entire life. And in this country, dialysis centers tend to be in regions where there's lots of people, so around urban centers. However, rural patients have higher rates of kidney failure than urban patients. And this is driven by higher rates of diabetes and hypertension and less access to early care. So they actually travel 30, 40, 50 miles to get to their dialysis appointment. And that's not once, that's three days every week for years and years and years, their whole life. So we have not solved the physical distance problem for chronic kidney disease. But we've documented it very well. Now, physical access failure is not just a rural story. So I want to make sure that's very clear. It's the patient who relies on public transport. They take the bus and you moved your clinic three miles away because the lease is cheaper. Now they can't get there. It's the patient who shows up to the office and they have a mobility issue. And yes, your building is ADA compliant on paper, but functionally it's not accessible at all. So they can't get in. Or it's the patients who can't afford the parking in a huge urban center. I'm literally driving down to the city tomorrow, and parking is ridiculous. That's all there is to it, right? Or they can't miss the wages, they can't miss not having their kids taken care of. There's all these different issues. So physical access failure is solved by redesigning where it happens. Not just where the building is, but where the touch point is. Is it at home, community centers, is it at a barbershop, a pharmacy, a church? The care delivery model has to move towards the patient before the patient can consistently come to an office. The second pillar pillar is financial. Oh yes. So it might backfire due to trust or some of these other pillars. So I don't think there's a straightforward answer to that one. So the next pillar is financial. This one seems obvious. Can they afford it? But it's not just the copay. I feel like the copay is the tip of that iceberg. It's on the top of the water line. But then there's everything below the water line that we don't really think about. All the submerged pieces, that half day of work that they have to miss, the child care, the $7 bus fare both ways, twice. The referral that comes after the labs with a specialty copay they didn't actually plan for. Being able to afford the medication for month one, and then not being able to do it for month two because rent's due. The follow-up appointment that was scheduled, and unfortunately, they had a child care issue. And the office issued them a no-show fee. So not only did they feel like if they went back, they'd have to pay that fee, but they didn't even really understand if they were allowed back. So financial access failure is really death by a thousand transactions. And most of those transactions are actually invisible to those in healthcare leadership because it's we measure the encounter. We don't measure all these other things, these indirect costs, because it doesn't necessarily impact the hospital system the same as it impacts the patient. And there's actually a consistent finding in medication adherence that even if you increase the cost of a medication by a very small amount of out-of-pocket, a modest threshold gets crossed, and all of a sudden, people are dropping their meds faster than you would expect. Faster than a rational economic model would predict. Patients aren't running cost-benefit analysis every time they go to the pharmacy. They're managing cascading financial pressure in real time for their whole life, not just that medication. I know I still haven't figured out how to use my damn SHA, and I can't get my money reimbursed. I have way too many degrees, and I still can't figure it out. And I'm sure many people in the same audience, depending on your insurance, are still struggling with reimbursement from your HSA. That's one simple thing. And everyone here has way too many degrees. Think about people who don't speak English, that don't read at the college level. They probably have no chance at all. The third pillar is cultural. Does the system actually speak to them? Not translated materials, though most systems are still really not doing that very well either. But does the clinical encounter feel like it's designed for someone like them? Or does it feel like they're completing a form built for someone completely different? The research literature on provider-patient concordance, and this is where the patient and the provider are of the same ethnicity, speak the same languages, all these different pieces where they're the same, shows that when we do have patients and doctors that look the same, talk the same, patients tell us more. They open up more, they're more compliant with medication suggestions. Chronic disease outcomes are measurably better. Not because of warmth, not because of bedside manner alone, but because trust is operational. And when people trust a clinical relationship, they use it differently. Our current physician workforce does not reflect the patient population that it serves. So it's not a diversity initiative, it's a clinical effectiveness issue. We are getting worse outcomes because we're not taking care of it. But cultural access failure is also historical. I know you've read all about this, I know you've seen this in all your lectures, but think about it. This Tuskegee syphilis trial happened in 1972. We had incorrect information in textbooks till the early 2000s around differences in physiology between races. And even today, black women are three to four times as likely to die in childbirth than anyone else, even when you control for economic stature, insurance coverage, etc. So it's not a patient engagement problem. It's a pattern the institution created over time, through documented policy and practice. Communities don't distrust medicine abstractly. They distrust it honestly, and they have receipts. So cultural access failure can be solved through lots of things. Hard things. Workforce investment through co-design with actual members of the community, treating community members as design partners and change agents from the beginning as well, not just research subjects. And it's solved very slowly. There's no accelerated pathway to rebuild the institutional negatives of the past. Digital health tools are almost uniformly most accessible to the patients who need them the least. The pattern's very consistent at every scale. A health system will invest in a digital tool, an engagement platform, you name it, they deploy it, and the utilization numbers are gonna be great. They look strong. And then someone actually disaggregates the data by income, by race, by language, geography, and the findings are the same every time. The platform's being used by patients who are already engaged, already making good decisions in medicine, already treating their conditions well, already comfortable with a portal. But the patients most at risk are not on the platform. They're being asked to download an app or do something else that they just don't trust. And I call this the digital access paradox because the more sophisticated a tool is, the narrower the population it actually reaches. And because performance is measured by engagement on the population that uses it, the gap's really invisible if you're in a healthcare delivery system on all your typical dashboards. You can have an excellent platform utilization number and actually be actively widening the access gap simultaneously. Now the COVID telehealth surge is the largest natural experiment we've done of this. And adoption exploded in 2020. When researchers disaggregated utilization, however, the pattern was stark. The highest adoption was among educated, ensured, urban, English-speaking, broadband connected patients. The population with the most to gain from telehealth had the lowest adoption rates. The technology was not the problem. The deployment design assumed a user that did not match the population. So digital access failure is solved by asking before you build, what technology infrastructure does this specific community have? Not what we wish they had, what they actually have. And sometimes the answer is a smartphone application. Sometimes it's a phone call. But you better not be calling me. Someone better be dead if you call my phone, right? Or maybe it's WhatsApp. But you have to ask the community. Sometimes it's an actual van on a Tuesday afternoon. The appropriate technology is not what you think it is, it's what the community says it is. And here's where I want to talk about trust, which I think is actually the most important pillar. And to my friend in the back row, this is it. Trust is the big one. This is where we're most reluctant to discuss the topic because it's the one we can't solve with a new shiny product. We can't buy this. Trust failure in healthcare is not irrational. It's not a communication gap. It's not a human literacy problem. It's a rational response to documented institutional behavior. Communities do not wake up one morning and decide to distrust medicine. They learn to distrust it, and we taught them to do that. There's even a very specific trust problem today in digital health that isn't being talked about nearly as much as it could be. When a patient uses a digital health tool, a wearable, remote patient monitoring, any application, they're generating data. And that data has financial value. It's used for product development, for research, analytics, population health, but it's also sold to third parties. And patients generating the data frequently have no idea the full scope of how it's being used. And they sign some random consent form that's built by the legal team for liability, not for actual comprehension. I'm sure we've all signed those same consent forms, and they don't even give them to us anymore. You have the little sign thing on your side of the desk, and they just tell you what it's for, and you just sign away. And communities with documented historical reasons to distrust institutions that claim to be helping them while extracting value from them, the data relationship is not an abstract concern. It's a familiar problem. We've been there before. So trust failure solved through radical transparency. In this case, about the data use, through community benefits agreements that actually ensure that those communities generating the data share in some of the value it creates. And having those people from those communities at the design table at the beginning, before the product exists, not just some random focus group afterward, not as a late-stage advisory board. A co-designer that genuinely has authority to change things. Did you have a question?

SPEAKER_04 26:31

How do I embrace that responsibility?

SPEAKER_00 26:37

So, from a policy perspective, when you're thinking about your AI policy governance, ultimately you need to decide how you're gonna be designing and what your institution thinks. If your institution believes that the communities that are impacted by the AI should be actually included from the beginning and should have the ability to make real decisions and change, those are policies that need to be written. Much of the governance that we have today in healthcare systems is around IT. And so we're using IT governance as a proxy, and the world of AI is much faster and it has a lot more problems. And so we're not actually building policies that make those things happen at the right time points. Also, in a press release, it looks great if you have a group of people of color in an advisory board, no matter when it is, and oftentimes these groups are put together for a press release and not actually to do anything good. Was there another question up top? So five pillars physical, financial, cultural, digital, trust. Every single one is a design choice. Someone made that choice or failed to make that choice. The patients who fall through those gaps did not fail. We failed to design for them. That distinction is not semantic. The entire argument is that. A researcher said to me, we built American healthcare from the inside out. We started with what was convenient for institutions, for health systems, for payers, for administrators, for clinicians. We built what was efficient for us, not for patients, for us. And then we handed this to patients and said, navigate it. And the result is that the healthcare landscape, where we have simultaneously, is some of the most advanced diagnostic technology in the world, in human history, and some of the most avoidable chronic conditions. We have AI systems that can read a diagnostic image with specialist level accuracy. And we have patients who will never make it to that appointment to use that imaging modality. It's not a technology problem. I know I've said that a hundred times. It's not a technology problem. The technology is fine. The technology is going to keep improving day after day, whether we fix the design or not. It's a design problem. We designed it for the wrong patient. Until we change that, no amount of additional technology is going to change who gets served. So, what does it look like to actually design for the right patient? Because I ultimately am very optimistic about our healthcare system and what we can actually do to change it. So I use three questions, and I ask these before anything gets built. A product ships, a new workflow is done, a new engagement strategy is put out, and I ask them again six months after we deployed two. Because oftentimes whatever worked on a whiteboard or in a conference room fails in real life with real patients. So I want to be honest about these questions. They're not sophisticated. An undergrad can understand these, but we're highly trained professionals in this room. And we're systematically not asking these questions. And the reason we're not asking them is that the answers can be very inconvenient. So question one, can they reach it? Not is the technology available, not does the website load. Can this specific patient in this specific life, in this specific context, actually use what you've put together? So how could you operationally test this? Well, take your workflow or your marketing campaign or your product and describe your median user. Not your target user, your median user. Now describe the population that you said you were building for. And why well enough to make a genuinely informed decision? And that doesn't mean they just sign it because they thought they had to, or sign it because they don't want the doctor to be mad at them. The operational test here is a little different. Have someone in that community read it to you aloud. Not a native English speaker from your organization, not someone who works at your hospital or company, someone from the community whose literacy level, whose medical vocabulary, whose relationship with healthcare reflects the actual population. And as they read through it, see where they pause, see where they reread, where they say, yes, I get that, or I don't know if this makes sense. This is where your solution is broken. And then question three. Will they trust it? Because I'm gonna come back to the trust thing every time. Will they trust it? And this is the question that most organizations skip completely. They just skip it. Because it's hard to measure, it's hard to get agreement on what that means. And the answers are very inconvenient. So, in this sense, how could we test this? Well, what's your community advisory structure? That's what we just talked about with that wonderful question from the middle. Does it have actual authority to make change? Not a board that meets twice a year to review decisions that have already been made. A structure where community members can say this feature or this flow conflicts with how my population interacts with healthcare. And then it gets treated as an actual design specification. You put it back into the product, the workflow, the campaign. It's not just for a press release, it's real change. So if your community advisory structure has no line to the people actually building it, it's not an advisory structure. It's a photograph for the annual report. And unfortunately, that happens every day. So let me show you what it looks like when we actually do this right. Not theoretically, not just in a book, in practice. In programs with real patients and measurable outcomes. This has been happening in lots of different places. So I'm going to give you two interesting examples. There's two health systems that have cracked this in different ways. Not by hiring more navigators or launching a social determination screening tool that no one actually uses, but redesign the foundation of the assumptions about what care looks like and how it gets delivered. So Hennepin Health is in Minneapolis, and they serve the patients that most health systems never want to touch. These are patients with serious mental illness, active substance use disorders, unstable housing, and often recent criminal justice involvement. Their ambulatory ICU has panels for their physicians of 100 to 150. Now the national primary care average is around 2300, to give you a bit of a different thought process there. And that number is not a staffing oversight, it's what it takes to actually know those patients. And each team includes not just a physician, but a social worker, a psychologist, and an addiction counselor, because behavioral health is not a referral for this population. It's the other half of the clinical encounter. They've also built cross-system data sharing between health care, criminal justice, and housing services. Because this population is different. The reason a patient lands in the ED is rarely medical. So this has actually allowed ED visits to drop by 9%. Inpatient emissions have dropped again by 3%. And outpatient visits have increased, which is actually the point. Another model is at CENMED. Now, this is a physician-owned primary care network for Medicaid, Medicare Advantage. And they've solved a very different version of the problem. The patients here are elderly, medically complex, and in that low to moderate income level. So a physician in this case has a panel of 350 to 450 patients. And they actually see their patients on average 13 times a year. That's 200 minutes a year. The average primary care number of minutes for most Americans is less than 15. So door-to-door transportation is included in their care because for elderly patients with multiple core morbidities, transportation is imperative for medical care. Hospitalizations at ChenMed run 50% lower than other organizations. Total cost of care is down almost 30%. The financial risk model here is full risk capitation. And I know you've learned all about that. So here, preventing hospitalizations is financially rational for them. And when the financial incentives align with the clinical mission, the outcomes follow. Now neither of these is a perfect template by any means. Hennepin requires cross-agency infrastructure that most healthcare systems are not willing or able to actually execute on. And Chen Men is very specific to Medicare Advantage. They haven't tried this model with the uninsured or Medicaid. And we all know that full risk capitation, when the margins are tight enough, it can create an opposite incentive where quality can slowly slide and care can be rationed. But both organizations prove the same thing. This is a design problem. The systems that reach the patients we never designed for are not staffed by more compassionate providers. They're built around different structural choices. Small panels, integrated behavioral health, financial incentives that reward keeping people well. Transportation that belongs in the care plan. So the patient we never designed for doesn't need a better app. They need a system that was actually built with them in mind. And this is what I call zero distance care. Not the elimination of geography, though that matters, the elimination of every barrier between the patient and the care they actually need. Physical, financial, cultural, digital, and trust. Zero. And here's what zero distance care is not. It's not community health workers deployed as a fallback when digital apps fail. It's not a low-tech alternative for populations without devices. It's not charity. It's not a charity model. It's design standard, the approach that produces outcomes. It's what we should have been building from the beginning. Zero distance design starts with the patient and not the shiny product. So before you open a design sprint, before you hire an engineer, before you write that policy or build out that marketing campaign, you have to actually talk to the patients. You have to talk to the community members. You got to ask them two things. First, what's your current relationship with healthcare? Not what do you want from it, not anything else, just what's your current relationship with it? Where did it break down? What was the last thing that made you decide the system was not worth your time? Then what already works in your life? What services do you trust? What institutions do you already use? Because the answer to that second question is often your deployment infrastructure. So if a community trusts the barbershop, well then the barbershop can be a community health touch point. If the church is it, well then that can be a healthcare access point. But depends on the population. Then and only then do you design backwards from what you learned. Technology is the last decision, not the first. And the appropriate technology is determined by the community's actual context. Sometimes it's a smartphone app, like we said. Sometimes you gotta call someone. And sometimes it's that van on Tuesday afternoons. So zero distance care is not cheap. Sometimes it doesn't fit very nicely into a 12-month-month budget cycle, but it's a version of healthcare access that can earn that name. So in this room, you study systems, and that's your professional training, your analytical lens. Why do systems fail? How can they be designed differently? What is the policy levers and where do they sit? That's what you do all day. I'm sure you had an assignment out last week. So I'm gonna say this directly to this audience. The problem I've been describing, you are uniquely positioned to address. And not just by research, not just by practice, but in a design sense. And the timing actually matters. The most consequential thing public health can do in the next decade is occupy the design table earlier. Not evaluate programs after they launch, not publish a paper in a big journal about why something failed. The design table. Before the product exists, before the workflow is decided, before the budget's actually allocated. And here's what I've learned in all my years of building on the tech side. Because I've been running these teams for a long time. The decisions that determine who gets served and what's in the product, they literally happen in the first three months. When someone writes that technical specification and they decide who the user is going to be, that's it. We have to be there. Because if the people designing don't understand the biases that could be included, or the population itself, they're going to make a lot of assumptions. So the question is not whether your research, your community expertise, your practice belongs in those rooms. It does. The question is how you're going to position yourself to be at the table before anything gets deployed, rather than in a big journal after it fails. And that's a professional strategy question. And it's one worth answering before you leave this building. So let me come back to 2021. Let me come back to my 90-minute drive. So we made it. We got vaccinated. And we were lucky. I had family in upstate New York. I had a car, I had flexibility. I want to be honest about that rather than skip over it because we're in a health access discussion here. But that same year, neighborhoods all across the United States, in this country, there were families that could not drive 90 miles, there were families who could not take the day, who found the phone tree. Ran out of data and gave up. They didn't stop because they didn't care. They stopped because the system had been telling them again and again, consistently and clearly, that it was not designed for them. And here's the thing I keep coming back to. She was right to stop. And it's not a sentence we say in healthcare settings, but it's the honest one. Given all the information she had, given the barrier she hit, given the three times she tried calling, the rational decision based on the evidence the system had provided to her was to stop. The system had communicated as clearly as any system can that it was not built for her at all. We taught her to expect that. We built the system that confirmed it. And then we used the resulting gaps in care as evidence that certain populations are difficult, not engaged, noncompliant. They're not noncompliant. They're making rational decisions based on the experience we gave them. So what if in the same city that year a community health worker knocked on doors, brought materials reviewed by community members, offered scheduling that took five minutes, face to face, provided a ride, same population, same year, different design. The technology didn't fail those families. The design did. And that means a different design can fix it. So here's what I think about with my own kids. I'm not worried about whether the technology they're going to encounter as adults, sorry, I got all these slides over this, as adults will be sophisticated enough at all. It's going to be. The technology will keep accelerating whether we fix the design or not. We're going to have genomic medicine. We're going to have precision medicine, even though it's been a four-letter word for 10 years. We're going to have precision diagnostics, AI therapeutic decision making, all these things at a level I can't fully even comprehend at this point. I'm worried about whether we will have fixed the design problem. Before they actually need care from a system that still doesn't know how to reach everybody who needs it. Because the patient we never designed for is not an abstraction at all. It's the mother who searched nearly a week and drove 90 minutes. It's the farmer who drove 40 miles to a clinic that didn't speak his language. They're in every zip code, every language, every insurance bracket, every level of digital fluency. They deserve a healthcare system built from them from the beginning. Not retrofitted, not approximated, not served by a workaround because we've messed it up the first time. Every person in this room is going to be making decisions about what gets built and what doesn't. Even the person asleep on the top row, even you. Every research design, every program evaluation, every policy recommendation, every budget allocation that you will make in your career is a choice about which patients will get care next year. So before any of those decisions get finalized, I want you to think about these three questions. Can they reach it? Will they understand it? Will they trust it? And if the answer is no, go back and redesign it. Don't just let it through. Redesign it. Because she did the math on whether to try one more time. She weighed the cost of the bus ride, missing work, the hold queue, the three times she had already been turned away. She decided the system was no longer worth the effort. She was right, based on every single piece of evidence we gave her. So we have to make it worth her while. So with that, I'm happy to take questions. Yes.

SPEAKER_03 50:01

But what doesn't give me a access to access on GI added?

SPEAKER_00 50:08

Absolutely. So I believe it's system design. And you've very nicely characterized the fact that we have a very expensive design that's not actually caring for the patients that we'd like to care for. And in other more single-payer systems around the world, we see them spending less money because they've designed it completely differently. And so this can apply to tech, it can apply to policy, it can apply to systems. And the systems are the things that need to be improved. Because just like you mentioned, we're going to keep spending tons of money and getting all the latest and greatest technology, but if it doesn't reach the patients and actually improve outcomes, what's the point? Good question. Yes.

unknown 50:54

Because there is a learning curve from the community to be specific table.

SPEAKER_04 50:59

There's also a learning curve for asking that.

SPEAKER_00 51:02

Yes. So we should have started decades ago. So the best thing we can do is start now. And sometimes these things can't be rushed because trust is something you have to earn. You can't buy it. So one of the things that someone can do is recognize what the population is that you serve. Who are the leaders in the that community? Are they willing to talk to you? And have them help you understand what might be the best ways to gain trust within that community. Because I'm not qualified to go into lots of communities because I don't have the expertise of that community. Doesn't matter how many letters I have after my name. But some people, community health workers, etc., we oftentimes treat them like volunteers. We don't pay them. We treat it as a leave behind. In a system that cares, you have to make it an actual position. You have to pay these people, and you have to have real goals to make it happen. I'm gonna go to this one first. So if your population uses WhatsApp, then you should be figuring out how to communicate with them in WhatsApp. If they're like my mom and are on Facebook 24-7, we figure out how to use Facebook. We oftentimes will put words like HIPAA as a big blocker for all sorts of solutions. Everything's figure outable. We can solve these problems. There's ways to make things secure in different ways. So if a community is on board with something, let's figure out how to make it work. And the other thing is recognizing that we speak a completely different language in healthcare than most humans do. And we forget about this because all our friends are in healthcare, because our colleagues in healthcare, because we talk about healthcare all day. But normal, reasonable people sometimes have no idea what the heck we're saying. And so when we're making communications, engagement, marketing, etc., we have to remember to bring those community members back in because they may have no idea what the heck you're talking about. Yes?

unknown 53:50

I'm wondering if since you're now experienced around the world, um, are there places where you've seen the system serving people better?

SPEAKER_02 54:00

And how much of that was there was also like racial and cultural influency, or were they able to serve people better in a similar scenario where people just be different of different cultures?

SPEAKER_00 54:15

I think that it might seem from the outside that a community or culture is homogenous, but the more alike everyone looks from the outside, the more different they are on the inside. So I'd suggest that any population that might seem from the outside as very uniform probably has a lot of nuance. The systems I've looked at, the systems I've worked with, everyone has some sort of heterogeneous aspects to their population, whether it's men and women, whether it's different racial groups, whatever it is. And the trick is they have to make sure that they understand what the population is they're serving. So I was on a call yesterday with someone from the VA, and I was talking about the five pillars, and to them, they didn't even think about the financial pillar because it's kind of taken care of in the VA. Whereas my other colleague, financials is the big deal. So you have to understand what the system is making important, what's already taken care of. And a lot of the semi-single-payer systems outside of the United States, they've already decided what's important. The whole country's culture believes in something different than we do in the US about how people should be cared for. It's their culture. And so I find that if you think about Americans and what the Americans in the whole United States, not this room, in the whole United States believe about caring for people and what people deserve or what people should have, it's different. So there's a big cultural aspect there that's not just about the different groups, but what the culture believes everyone should have.