Kristen Miller: A Guiding MedStar for the Patient’s Journey
Kristen Miller is a healthcare delivery researcher with nearly two decades of experience working with MedStar Health. Her work focuses on improving how care is delivered across all phases of the patient journey, from preparation before appointments to follow-up after visits. Unlike traditional bench science, her research emphasizes real-world tools and systems designed to help providers and patients make safer, more informed decisions.
She emphasizes the importance of tools tailored to various stages of care: pre-encounter preparation, in-visit communication, and post-visit follow-up. Her team is working on simple, accessible aids, such as cue cards and checklists, to help patients describe symptoms more effectively. She also highlighted the widespread issue of diagnostic errors, noting that one in 20 people will experience one in primary care annually, and that all patients are likely to face at least one in their lifetime.
Kristen Miller is a fantastic presenter. She has appeared live at numerous conferences, including PULSE’s Pulse's Family-Centered Patient Advocacy Training New York this past March.
Resources Mentioned in the Podcast:
Be the Expert on You (tool from MedStar and AHRQ)
PatientPartneredDCE.org
ECRI Top 10 Patient Safety Concerns
ChatGPT (mentioned in context of AI usage)
Timestamps of Important Moments in the Podcast:
00:51 Kristen explains her focus on healthcare delivery research
03:27 Description of clinical decision support tools for providers
06:14 The evolution of risk tools from provider to patient use
08:28 Kristen discusses medical gaslighting and patient dismissal
12:55 Diagnostic errors in primary care and patient lifetime risk
14:56 “Be the Expert on You” tool and symptom description training
17:32 Adapting patient tools to different needs and health literacy
21:31 Empowerment during the encounter, not just preparation
25:14 Discussion on types of AI tools and patient trust
31:53 Diagnostic safety as a core focus of medical care
35:03 Misdiagnosis of cardiovascular disease in women
38:47 Challenges of provider pattern recognition and AI comparison
39:56 PatientPartneredDCE.org and ways for advocates to participate
41:40 From patient education to full diagnostic empowerment
Host
Kristen Miller, thank you so much for being on the podcast.
Kristen Miller
Yeah, thanks for having me.
Host
Love it. So, you know, we get a fair amount of opportunity talking to advocates. We don't talk to people on the research side all that often, which is, which I think, you know, the research informs the way hospitals work, the way patients kind of interact with the medical system. And so it's kind of in many ways the root of how we initiate and enact change in the medical space.
Talk to me about how long you've been in the research world on the medical side. You work with MedStar, for those of people who aren't familiar with Kristen Miller, and what you've learned kind of over the years, the big takeaways of doing research.
Kristen Miller
Sure. Yeah, I personally have been doing this work maybe close to two decades. I do think the research that I and my colleagues and peers do, it is unique. It's what we call healthcare delivery research. And so our goal is to improve the way that we deliver healthcare. And that can be anything from designing interventions, improving the tools that providers use, that patients use to do a number of things, make it easier for...
providers to do their job and follow evidence-based guidelines, make it easier for patients, ensure that they're having a good experience, that their care is safe. So I do think it's sort of a unique aspect of research. It's not the traditional bench science that you might think of, but really sort of that practical applied, how do we make a difference in the way that we're delivering care today and tomorrow.
Host
And so when you do this research, how are you actually conducting it? If it is engaging with patient experiences, are you doing kind of longitudinal studies with specific patients and different disorders or different hospitals? What's the actual research look like?
Kristen Miller
Sure, I think a lot of different methods, primarily we do more qualitative research. And so I think at the root of this is trying to understand what's the current state, what is the experience that everyone is having, where are the gaps where we can make the biggest difference. so anything from interviews, focus groups, surveys, to more observational ethnographic research. And then as you mentioned, more of the longitudinal. So.
We want to understand where the gaps are. want to develop interventions, but then we also want to deploy them and evaluate them. I think in the work that we do, we certainly want to know if something is working, but we also want to know in what context is it working and who is it working for. So as we develop tools for patients, we want to appreciate.
Is this a tool that works for every patient? Does it only work for a subset of patients? Does it work only for someone with a new diagnosis or does it work for someone with a chronic condition? And then we iterate and we optimize to make sure that those tools really have sustainability, but also scale. You mentioned MedStar, large healthcare system on the East Coast. We do a lot of work implementation-wise at MedStar, but we also want to make sure that the tools that we're using can be.
more broadly applied. Can other healthcare systems across the country and even across the world use some of these same interventions that we've developed?
Host
And when you say tools, how broad of a spectrum are we talking when it comes to tools? Because it's a very generic term, obviously.
Kristen Miller
Yeah, great question. It's really any number of things. The research that I started was in what we call clinical decision support. And so those are primarily tools that are provider facing physicians, nurses. They are generally tools that work inside the electronic health record. And so maybe it's pulling things like your symptoms, your previous diagnoses, it's alerting the provider that they should run this certain test or they should think about the certain diagnosis, for example.
That's where my work had started. And I think we realized that work is critically important. Providers have a ton of production pressures. They're trying to do a lot, you know, in primary care and a 15 minute visit. We want to support them and give them sort of the best evidence and maybe alert them to things that they otherwise might overlook. But there's...
Host
So just to slow this down a little bit, so I'm making sure I understand. you go see a doctor, the doctor has their EHR, their electronic health record of you on a monitor typically as they're visiting with you or sometimes right before they visit you. And so there's a tool within that software that'll say, Hey, here are some of the past diagnosis of this patient. Maybe you should check for this. Maybe you should check for this.
and kind of suggestions or options so that they're more efficient so things don't get left on the table that normally might get unsaid because there's just so much production pressure. They gotta see so many patients an hour. So the tool kind of lives within that software as an example. That's what you were talking about.
Kristen Miller
Yep, exactly that. And I can give you one concrete example. Perhaps the most successful thing that we have built to date on the provider side is a cardiovascular risk calculator. And so it pulls things like a history of smoking and your cholesterol and your blood pressure. And if you're on aspirin therapy, a number of different things that would be difficult for a provider to sort of calculate in their mind. And it presents a 10 year
risk of something adverse cardiovascular related happening. So over the next 10 years, how likely are you to have a stroke, heart attack, heart disease?
Host
How does it express that risk? it a percentage? Does it say you have a 70 % chance or is it like a great, I'm just curious. So like, does it act?
Kristen Miller
Yeah. So that one is, it is a percent, but it also sort of puts you in a category because generally humans are not good at numbers. We hear 10 % and some people might think that's extremely high and others might think I'm good to go. And so it helps to not only quantify the risk, but also put you in a category. Are you low risk? Are you borderline, intermediate, high risk?
Host
Right, right, and that's more useful for a broader audience, both probably both on the provider and the patient side, I would guess.
Kristen Miller
Exactly that. So we built the tool initially for providers. We called it the risk estimator. And then again, we thought this is really important information for the provider to know. It's really important for the patient to know as well. I don't know that anyone's ever told me my cardiovascular risk. And so we created a second visualization that we called the risk educator. And that was meant for patients to be able to understand their own risk. So in the example you gave, you're in the primary care encounter, the doctor's looking at the EHR.
Host (06:29.4)
Right.
Kristen Miller
And what we had envisioned was the provider actually maybe turns their screens to you and invites you in to the conversation and says, let me show you your risk. Let me show you why this is your risk. And here's what we can do about it. So we talk about different things, modifiable risk behaviors. Should you quit smoking? Should you focus on your diet? We can't make you older or younger. We can't change your race, but there's certain things that we can do. And in real time, the provider can show you.
If you quit smoking, here's how your risk will change. If we lower your cholesterol, here's how your risk can change. And I think of myself as a patient, that's way more powerful than somebody telling me your risk is 15 and we need to get it to 10 and like, let's think about it, right? You're giving me actionable things that I understand that I know I can do something about.
Host
Well, and from just a purely psychological standpoint, if there are things you can do, you're also empowering the patient with hope. And if for some reason they're in a dire situation where there isn't much hope, at least they can start working on the acceptance and understanding quality of life or whatnot, you know, for, where they're at, and, normalizing it. And which leads me to like a bigger, broader question with you.
Kristen Miller
Yeah.
Host
and it sounds like that not only was that a useful tool in terms of making things run smoothly, but you're really improving the patient experience with that process because they feel like they're partnering with their physician. mean, every time a physician turns the screen to show me the X-ray and explain things or a chart, I feel much more educated and empowered. And I would assume I'm not unique in that sense.
Kristen Miller
Sure, yeah. And if you look at where the focus has been more recently, things like the ECRI Top 10 Patient Safety Concerns for 2025, other reports, we talk about medical gas lighting and dismissal of patient caregiver symptoms. And so I think there's sort of that misunderstanding. Nobody wants to feel sort of less than in their own primary care encounter or their own hospital admission or whatever it is where people are using
medical jargon, this stuff is already complicated. You're already at a disadvantage being a patient just in general. And now you're trying to appreciate the risk of something bad happening or a new disease condition that you have. And so what are the things that we can do to make that easier? How do we assess things like health literacy? How do we make sure that you leave that visit knowing more than you did when you came in, but also being empowered yet to do something about it and recognizing.
what is within your power. It's not just conversations about surgery or medication. It's really about understanding and accepting this is the new diagnosis that I have and here's what happens next.
Host
Yeah. And when it's done poorly, everyone notices it's one of those things when it's done well, you kind of flies under the right radar and says, okay, that that was a doctor visit. But when it's done for, I remember 20 years ago, you know, when I was kind of a young adult and just getting used to it and patient experience was still really rocky in a lot of places. And I got some blood work done. And then I got the the just as an example of a bad patient experience, I got a phone number to call
Kristen Miller
Right.
Host
where it was like a tele-automated thing where they gave me the results and it said, your blood pressure is high. This means you're at risk of many things. And that was the end. There was no follow-up number to call. Like it was, you know, I just kind of got my own doctor who wasn't a pediatrician and I felt so lost. I'm like, great. So I'm dying and I don't know what to do about it. And it was, it was incredibly frustrating. So it's good to see that there's more focus over the last two decades since you've
Kristen Miller
Yeah.
Host
kind of come into the research part of it where people are realizing, hmm, we gotta make sure that people feel educated and empowered throughout this process.
Kristen Miller
Yeah, and I think you had mentioned tools, right? So I gave you an example of a provider-focused tool, and that's great. And we need to put more tools in the hands of everyone, including patients as well. And so in a lot of this early research and talking to patients and talking to caregivers and advocates, we have organized the way that we're thinking about this as sort of the pre-encounter. How do you know that the symptoms that you are experiencing
Host
Yes.
Kristen Miller
warrant you to need care and what kind of care you should engage in. That's sort of the pre-visit. And then even if you don't have access issues and you get something scheduled, how do you prepare yourself to have that conversation? And again, I think we've all felt this of, I can't forget to ask the doctor about this. does the doctor need to know about this thing that I'm experiencing? And so all of this sort of pre-encounter.
work and developing tools to help providers or help patients prepare for those visits. Then separately, there's what happens in the encounter. And I think that's where we often think about where improvements can be made. We're co-located in a space, whether it's physically or telehealth, we're trying to almost like detective work, figure out what is wrong with you. Why might you be having these symptoms? And what is the answer and what do we do about it?
And then we're starting to engage on what happens after the visit, because even if the visit goes well, if the patient leaves and doesn't really appreciate what happens next, was I referred to a specialist and for what and how quickly do I need to schedule that visit? Or do I need to be making lifestyle changes? Or do I need to start taking this medication? So really thinking about sort of the pre-encounter.
We're developing tools for that. What can we do to improve the actual encounter itself? We've been focused on primary care because that's where the majority of diagnostic errors can happen. It's estimated that one in 20 people will experience a diagnostic error in primary care this year, but then also the follow-up of what happens after.
Host
Justin, Justin primary care, one in 20, Justin primary care. Wow. That's stack.
Kristen Miller
Fist in primary care. And the bigger statistic is that this will happen to everyone. At least everyone will experience a diagnostic error once in their lifetime. And that doesn't mean it needs to be an egregious event. We often see very delayed diagnoses, often in cancer and other disease conditions. But it could be a misdiagnosis, a wrong diagnosis, or it could be something that you just don't understand. It wasn't explained to you.
in a way that you left the office knowing what hypertension is or knowing what diabetes is. Even if we get the diagnosis right, it needs to be communicated appropriately to the patient and the family.
Host
So, I mean, just to put it, I do like numbers even if I can't wrap my head around it, but what you're saying is this year, according to the numbers, if I go to my primary care physician, I have a 5 % chance of some form of misdiagnosis in that primary care visit.
Kristen Miller
And I think that you are correct. I think that may even be an underestimation because it's very difficult to measure something that we didn't get. Right? If on the research side, I'm trying to say, did we miss a diagnosis? Did we have the wrong diagnosis? It might be years before we actually get the right diagnosis. And so from like a pure quantitative research standpoint, it's very difficult to estimate what the true risk is.
Host
Right. But that's the conservative estimate. That's the conservative estimate. Wow. Which makes you appreciate it a little more as a patient, like, do your, do your part to make this likely to go in your favor. What kind of tools are you working on or have you worked on for the patient because they don't have access to the EHR. is it, are there forms that you give them in the clipboard when they kind of come in? Are there like different charts? mean, you know, something like as simple as a pain scale to kind of help them understand
Kristen Miller
Yep.
Kristen Miller
Yeah.
Host
how to describe their pain. that what you're talking about?
Kristen Miller
Yeah, so one tool that's available right now, you can find it on the MedStar website, but also the Agency for Healthcare Research and Quality is called Be the Expert on You. And that is really an agenda setting, goal setting tool. When I had mentioned how do you prepare for the visit is really thinking about all of the things that you need the provider to know about you. What are those symptoms? What medications are you on? What questions do you have for the physician?
There are a number of agenda setting tools like that, regardless of what tool it is, something structured to really help you prepare for that visit. And also recognizing that you are the expert of yourself, right? You're the ultimate arbitrator of your own health and your own diagnosis. You know more than anybody else about what you're experiencing. And so how can we articulate that to the care team in the most actionable way? Related to that one.
Host
How is that? I'm sorry, go ahead and continue. didn't mean to interrupt.
Kristen Miller
No, no, we're just so excited about all of these different things that we're building. One other thing that we've just started working on just as a simple example, no one has ever taught me how to describe my symptoms. Just things that we don't really think about. And you can imagine a patient who shows up in the emergency room or at their primary care physician or at the urgent care, the difference between saying my back is killing me, which
is alluding to a certain symptom, but it's not really providing a lot of information. It's very different than if you helped someone learn how to structure a discussion of their symptoms to say, this is the type of pain I've been having, stabbing pain, it's been going on for three days, it gets better when I lay down, it's worse in the morning. Right now, all of a sudden, you've given the provider a lot more information that's gonna help them get to that ultimate answer of what might be going on with you. So.
The tools can range from very complex risk calculators to very simple cue cards. We envision, what if I handed everyone in line at the emergency room a little pocket-sized card that helped them structure their symptoms when they got to the front of the line? What would be able to get to the right answer quicker, faster, right? In a way that really improves not only the efficiency but the safety for the patient as well.
Host
And so you kind of answered the question I was going to interrupt you with, which is how do you distribute these tools? And one example you're using right there is what if we can get it on a card or even a poster or something that everyone in the ER or the urgent care unit can see before they actually talk to a physician or even an intake coordinator of some sort?
Kristen Miller
Yeah, and I think the way that we think about this, especially what we've learned from patients, is there is no one size fits all. There's nothing that's gonna work for everyone. We think about things like health literacy and the preferences of an older adult, for example, compared to maybe more tech savvy teenager. Everyone wants something different. And so let's create these tools in different ways that are accessible. So for some, may be...
a screening tool that you use when you are getting, you you're registering for an upcoming appointment. For others, yeah, it might be that piece of paper. It might be a poster on the wall that you see right at the last minute. And the more that we can disseminate these tools in different ways to different audiences, I think we're gonna see more success because everyone might need something a little bit different.
Host
when we talk about the patient experience, how are you measuring that? Is that like a satisfaction kind of thing?
Kristen Miller
Yeah, another great question and another one that's really difficult. again, I think we put multiple layers on everything so that we can really try to understand what's going on. When we think about the evaluation, there are lots of different endpoints that we're trying to understand. So there may be process-related things. Did the patient get the right diagnosis? Did they get the right antibiotic for their infection? Right, like there's...
certain things that happen in the clinical process that I might be looking for. There are patient experience outcomes that are more maybe patient driven, a conversation with the patient, how did that encounter just go? The survey that you might get a few weeks after your visit, that's sort of the more, yeah, HCAHPS, Prescaini, a number of different companies that facilitate those. And we can look for indications of...
the patient experience and specifically diagnostic error. We could talk to providers. We can look at the electronic health record and the documentation that was made. And what we try to do is triangulate all of that data to get to an answer or use surrogates because it's rare that when you're leaving the visit, is there a researcher right there with a clipboard saying, Jon, how did that go for you? Like, do you feel like you got everything that you need?
And so, I mean, to that point, can we put better systems in place to be able to capture that? And so that it's not months away where somebody's saying, you know what, I didn't understand what that new diagnosis was. I'm still confused. Can we catch it sort of right off the bat?
Host
Yeah, that's interesting. And you're starting, I'm starting to see just how difficult some of this research is to actually collect and analyze because obtaining the data becomes quite complicated.
So in 20 years, might only get so many things done, especially if it's a longitudinal thing where you're kind of tracking patient studies over time. That's crazy. You you mentioned preparing yourself for the physician visit, for that encounter, as you called it. And it's so interesting because a lot of our advocates, Karen Mercerow comes to mind, she has something she calls the key. And one of the things she does as the advocate for the patients is she does this in detail.
as do many advocates we've spoken to here on the podcast, so that when you go in and you have your 10 or 15 minutes with the physician, you actually do know what to say, don't leave anything out, at least not hugely big, that would make a huge difference so that you can benefit from it. I like, obviously in an ideal world, every patient has that kind of education, but if not,
How do we get 10 % more, 20 % more? Is that really the goal is to try and figure out how to get as large of a population, as much adoption of this kind of mentality as possible so that it just helps everybody?
Kristen Miller
Yeah, and again, I think everyone needs something different. The preparedness part certainly is important. We think about sort of the levels though, right? So the sprained ankle is very different than a cancer diagnosis, right? something more acute versus something more chronic, something less severe, more severe. And there's like tons of different levels, whether you're taking care of your child or you're an adult.
you know, taking care of your parents, yourself. I think we want something for everyone, but we want it to be what they need. And what we found is that everyone needs something a little bit different. You mentioned the preparedness part. I think that's really important. I think what we found though sits more in the encounter itself and helping patients and advocates actually do the advocating. And how do you...
feel empowered to ask the question, to use your own voice, to know that you can ask questions, right? And not feel like, I don't know, there's a power dynamic or you feel dismissed and so you're not willing to ask questions. And so some of the work that we've been doing is really focused on the empowerment in the encounter, right? If the provider continues to interrupt you, if they're being dismissive, if they're using...
biased and stigmatizing language, what are you supposed to do in that encounter to really feel empowered and speak up? And so that's sort of a very different level of tool where we can prepare for the visit all day. But we also want people to feel comfortable in that moment. And I think that's work that needs to be focused on the patient side, but the provider side as well. So we spend a lot of time with medical students and residents and
I have seen some pretty significant changes in the way that we're training providers to recognize their own bias, to recognize that the patient is an expert on their own body and sort of change the dynamic of what that traditional encounter used to look like.
Host
Yeah, I mean, I've had family members who've said, you know, something as seemingly innocuous as an eye roll from the doctor will get them to just completely shut down. It's like, well, I was reading on WebMD or Google, then you get this eye roll from your doctor. They're like, all right, I'm just not going to say anything else because now I feel like an idiot. You know, and it's like that. really does play a huge role in just how comfortable you are and having an honest conversation about things.
Kristen Miller
Yeah.
Kristen Miller
There's a great clip from one of the medical television shows where the provider comes in and the patient says, I think whatever it is, I think I'm having this event and the provider says, and where did you go to med school? mean, to feel so dismissed that how dare you even bring forward a potential diagnosis, right? That you think, I do think that times have changed and folks are using, whether it's WebMD or chat GPT, there's so many more tools.
Host
Right.
Kristen Miller
that patients and caregivers and advocates can use to, I wouldn't say get to the final diagnosis, but to get some good ideas, right? And start to think about what your diagnosis might be and start to advocate a little bit, learn a little bit, educate, you know, obviously with trustworthy sources, but that definitely changes the encounter then if you're going in, you know, with your own ideas and your own information.
Host
Since you brought up chat GPT and you work on tools, are you guys seeing more adoption or interest in the AI space when it comes to assisting patients and providers and kind of what's your own personal attitude towards that?
Kristen Miller
Yeah, we definitely are. think I'll sort of roll back the conversation a little bit just to say AI means a lot of things. And so we are doing a lot of work with patients to understand their perceptions of AI. But it became clear to us that people are using AI as sort of a blanket technology. And it's important that we distinguish the different types of AI that we're using in health care. So for example,
we are using ambient digital scribes now at MedStar. So that is a tool that records the encounter, a listening device, for example, right? And then it creates the provider note. Our goal is that it can create a patient note that really helps, you know, break down some of that medical terminology, explain, you know, a new diagnosis in a way that you can understand. That is a very different AI tool than, for example, a radiology AI tool, which is looking at an image
and helping the provider find something abnormal, which is different than we talk about chat GPT and a provider saying, hey, my patient is experiencing these symptoms. What might the diagnosis be? And there's at least a dozen different AI use cases that we have found. And there is value in all of them, but there's safety risks as well. And so trying to understand how do we deploy these in a way where we appreciate
safety and fairness and validity and like all of these other outcome measures.
Host
And privacy too. Yeah. It's interesting. The two use cases or you mentioned three, two kind of were in the same camp, but one is an efficiency tool. And I've even had the experience where I go into urgent care and the doctor takes his phone out or what looks like a phone. might be a hospital phone, you know, and says, I'm recording this to you. I have your permission. It just helps me take notes later. And it is an AI tool. And so you're like, okay, well, that's just helping things speed along. And actually I might even get a more accurate record.
from the patient perspective of what happened in the room, especially because I think more and more people are seeing tools like that incorporated in Zoom and in other places in their lives. So it's more familiar versus a diagnostic tool. Like you mentioned radiology where it's like, I'm trusting the robot quote unquote to like, tell me what's wrong with me. That's a lot scarier. And so adoption might be a little bit slower in some areas than other, but it's.
I think, you know, if it saves lives, everybody's for it. It's just a question of like, where's the risk? Where's the challenge on making it fair? But it's a fascinating feel. Are you incorporating it with some of the research you're doing already?
Kristen Miller
and
Kristen Miller
We are, and we were exploring the different diagnostic use cases in these different AI scenarios. We just published a paper on this where we talk about, we brought in 17 patients for some activities where we went through those scenarios. And what we wanted to know was their preferences about how we communicate this. And again, we think about the deployment and the safety of it, but there's...
huge implications for the transparency of even letting patients know that this is happening. How do we explain that to patients? When do we explain it to them? What do they actually want to know? And not to dismiss patients that say, like, I want to know the underlying algorithm, right? I want to know how valid this tool is. Whereas we had others that said, like, I just want to know, right? I don't want to feel like you're not sharing this information with me.
But it's interesting that you, the way that you frame the scenarios, because I think the biggest takeaway was patients were okay with it if it was a supplemental tool, right? If it was anything that providers were using that made things a little bit easier for them, gave them a second set of eyes, supported their decision-making, they were not okay with anything that would replace the provider. And so when we would give examples to say there's this algorithm, it can predict
your risk of something, there's still a recognition of the clinical expertise and the clinical gestalt. I want the provider also looking at it, right? I don't want AI to be the sole tool that's being used. And now we're taking those to create guidelines that we will put forward sort of best practice. Here's how you should communicate this to patients in a way that values their, that relationship, right? Doesn't undermine the trust it keeps with the transparency.
Host
Sure.
Kristen Miller
And that may seem like a simple thing, like, just put it in an informed consent. I let them know that we were doing it, but it really should be a conversation about why is this tool helpful? What is it doing? What happens to my data? Right? The listening example is a great one. You're recording my primary care encounter. You're recording the words I'm saying to you. Where's that data going? Do I feel like it's going to end up on the internet somewhere? And I didn't allow that. So a lot of different dimensions that we think about.
Host
Right, sure.
Host
And it's, and there's really the way I kind of, in my head, at least as someone who's not an MD or has spent any time in the hospital setting as a provider, there's really, I broke it down into three buckets just to help myself wrap my head around it where there's efficiency tools, right? So it's, you know, things like data entry. I think, like you said, most people are like, okay, whatever, you know, it's like, sure, maybe there's a question of privacy, but you know, if this is going to help you help more patients and not type up things for an hour.
I'm all for it. Just as a simple example, then there's the diagnostic tools where they don't want the provider replaced. And then even further, there's the treatment tools. So down the line, having like a robot start your IV or even like an autonomous surgeon bot making a decision in the OR. mean, that's the scariest one for most people. But there's huge spectrum there and it's fascinating to see. I also think just like autonomous driving, like
it gets more and more commonplace. People trust it more and more. I don't even look over my shoulder. use the backup camera. Whereas the first time you do it, you're like, nope, not trusting it. Right. So it's, it is, it's kind of interesting to see how the world's going to evolve and having people like you track that level of trust and concern is hugely valuable, hugely, you know, both to the patients and to the providers.
Kristen Miller
Yeah, definitely.
Host
Which brings me, know, since we talked about diagnostic, we don't talk a lot about diagnostic safety. And I know you and I touched on that in an initial call we had. Can you talk a little bit more about how that's come up in your research?
Kristen Miller
I would love to. So you're right. I don't think the term diagnostic safety, diagnostic excellence, the diagnostic process, that's not a way, at least that I was ever taught about medicine. When in fact, it's arguably the most important part, right? When you, it's, there are people that just go for their annual checkup every year, right? And might not have a problem. It's more likely that everyone has some sort of symptom. And again, whether it's something very simple, you have a cold, you have...
I don't know, the flu, something that's happening acutely to something much more serious, you're often going to the hospital or to your doctor to figure out what is this, what's wrong with me. And we think about that as the diagnostic process and it's everything from you experience a symptom and you recognize that you need care all the way through to those treatment decisions that happen at the end. But there's different stages of accessing the healthcare system.
information gathering where the providers may be asking you questions, they're running lab tests, they're hypothesizing what the diagnosis might be. Once there is a diagnosis made, it's communicated, right? And then you continue down that journey. I think what's interesting for me is that this does impact everyone. And what we want is that you get the right diagnosis, that you get it quickly, and that you get it communicated to you in a way that you really understand. So the levels we look at are accuracy,
timeliness and then communicated. And we have tons of examples where this went really well. And we have tons of examples where it didn't, where it was the wrong diagnosis. Maybe you had two things wrong and we only picked up on one of them, or we got it right, but it took us a really long time. We did the Imagen study. We found that there was a lung nodule. We didn't communicate that with you for a year or two. There's lots of different ways that...
this can go awry. And so our goal is to make sure that we get it right every time. If you get the diagnosis wrong, everything that follows is wrong, right? So this is really sort of the starting point of your care journey.
Host
And I think that the old kind of, the way I think people can think about this very clearly is in terms of like cancer, where someone says you didn't catch it early enough, right? It makes a huge difference to the possibility of going cancer free or not, or taking certain treatment options on and off the table. That's obviously a very dire example, but it kind of.
puts the framing of why it's so important kind of right on the table.
Kristen Miller
It does. And I think, know, I've come to appreciate where the failures can happen. The example that you just gave were sort of part way into that journey, right? But appreciating, we are getting it wrong at the very beginning quite often. And I think the first time I learned to appreciate that was a project that we had about misdiagnosis of cardiovascular disease in women. I'd like to think I'm pretty smart and I'm well educated and I'm around healthcare people. I had no idea that women
Host
Thank you.
Kristen Miller
don't present with crushing chest pain because that's what we've heard our whole lives, right? A heart attack feels like an elephant is sitting on your chest. That's all that we know. When in fact women present very differently than men, their heart attack symptoms can be jaw pain, it can be nausea, it can be all of these, right? All these things around your heart. Yeah.
Host
I'm blowing my mind. I've never heard of that before. That's crazy. And I bet some providers will dismiss it as TMJ or something if they, that's crazy.
Kristen Miller
Or a patient might think this is just indigestion and I'll be fine come Monday, right? And so I don't need to seek care. And so we're failing people at the onset when we're not even explaining to them what to be looking for, what to be worried about, what's a symptom of a disease versus something that maybe will resolve itself on its own. And so that's where we can fail at even accessing care is making people aware.
Host
huh.
Host
Yeah. You know, I'm starting just the way you presented that one example makes me appreciate just how difficult the physician's job is. Cause if a female patient comes in, let's say I do know this right as a physician and as a cardiovascular, you know, expert or even as a primary care, right. And someone comes in and says, I'm having jaw pain. On one hand, I have to now ask questions to say, okay, let's rule out cardiovascular issue. Cause
On the other hand, I don't want to tell them, are you having a heart attack and scare them into one. So it's a really kind of fine line to walk with where you also don't want to dismiss them so they don't tell you other symptoms. it's, and I would assume that's where tools become really incredibly important because it's like, well, how do I proceed in this instance?
Kristen Miller
Yeah. Well, so my training is human factors engineering, systems engineering. We're looking at the entire work system. We're taught never to blame anyone, right? We look for these errors that are sort of systematic to the process. And I have learned there are more than 10,000 symptoms, no, more than 10,000 different diseases, but there's roughly 200 signs and symptoms.
And so a provider is taking this information that you're giving them, right? The jaw pain, the nausea, the fever, and it's, you could just, just the math and the odds of that, right? To 10,000 different options from a handful of signs and symptoms. It's not an easy logic, right? If this, then that, right? If the patient says back pain and fever, it must be this. It's really complicated. And so in no way are we blaming the...
clinical team to say you should have gotten this right because it is really complicated. And so how do we get closer to the right answer? And again, is it that they have all of the information, that they know your history, they know all the signs and symptoms? Is it those predictive algorithms that are narrowing down from the 10,000 to a top differential of three different options? What are all the things that sort of increase the likelihood that we're getting the diagnosis right?
Host
Yeah. And sometimes those even look like magic in the, in the diagnosis. No, I'm serious. I've seen it many times with friends and family where it's like, I have my pain here and the doctor says, well, raise your arm. And suddenly the pain's worse than like, now I know what's wrong. Or when I went in and I had some pain and they're like, well, maybe kidney stones walk across the room on your tip toes and the pain got infinitely worse. And they're like, yeah, it's probably kid. Let's get you an MRI. It's like, what did the tippy toes do? And it's it, but it's like,
Kristen Miller
Yeah.
Host
Being able to drill that down because there are so few symptoms for so many different diagnoses, it's a crazy gridlock of a matrix of options you have to kind of go through.
Kristen Miller
And that's that clinical gestalt, right? That's provider who says, I've seen this before, right? And they're using diagnostic reasoning, but they're also just pattern matching in their mind. I remember, right? I've seen this case and I asked them to walk on their tippy toes and here's what we found, right? Or I had another, you know, more experienced clinician who told me to always ask them to raise their arm. And that sort of the, takes time to get there, right? These are things that are learned over time and through experiences and seeing them with patients.
And so how do we, again, how do we expedite to get there so the provider knows to ask that question to get that more information?
Host
And I think that's partly why people are less trusting of AI with diagnoses, because they've seen it in their lifetime. there's just an intuition there that a computer screen won't think to ask. It's like, when's chat GPT say, hey, go walk on your tippy toes, let me know how it feels. You know, it's an interesting kind of world that we're kind of evolving into. If people want to participate in some of these studies, the website that you've mentioned to me before, just to make sure I have it right, is Patient Partnered.
DCE org is that right?
Kristen Miller
It is, yep. So this is one of our diagnostic projects that's funded by the Agency for Healthcare Research and Quality. We're one of 10 national diagnostic centers of excellence, and ours is focused on patients as partners. And so we try to keep this updated with new research studies, posters that we've presented at conferences, the things that we're reading, even if we didn't help to write them, and then opportunities for patients to get involved. So if there are...
specific studies where we're looking for patients with a specific diagnosis or it could just be a role. And we're wrapping up a study right now where we did talk to advocates and a lot of them, GNA advocates about their diagnostic experience. What tools have they used? How would they imagine this going in the inpatient setting? So tons of opportunity for partnership at multiple levels.
Host
So it sounds like even advocates might be welcomed in some of these studies as participants.
Kristen Miller
as participants and as experts with that same lived experience. I, you know, when we talk to advocates and caregivers, they're looking at it with a different lens. And I don't think it's, it's the expertise that they bring as advocates, but it's also as that second set of eyes, right? They've seen what's worked well when the patient responded to a framing of a question, right? And they've seen what didn't work well, like the eye rolling that, you know, sort of made someone retreat a little bit into their shell. So,
We're trying to learn from everyone and bring all of that expertise, surface it all together to again, like create solutions that everyone can use.
Host
Is there anything we're not talking about or that you wish people were talking about more in the medical space?
Kristen Miller
I think for me, it's just trying to raise awareness about this idea of diagnostic safety and patient engagement. I think in many ways, we're still at this sort of early education awareness. You have a role, you can speak up, you can ask questions. For many patients, we're still there. Ultimately, we want to get to a place where
the diagnostic process is a true partnership where the patient is considered part of the care team in the same way that the nurse and the radiologist and the pharmacist, right? We all have equal roles in this process. And that's really moving us from this idea of education to engagement and then ultimately to empowerment. We want patients to feel empowered to ask questions, right? To bring forward their own ideas, to participate in shared decision-making, right? Other concepts where
the patient really is at the center of the care process and not just a participant in their own care journey.
Host
Kristen Miller, thank you so much.
Kristen Miller
Yeah, thanks for having me. It's been a privilege and a pleasure and I hope that we can work with more of the audience and really make a difference in this space.
Host
Me too.
