Designing for Health: Interview with Shreya Shah, MD [Podcast]

Show Notes:

[00:00] Intros

[00:26] Background and career path

[01:00] Path into clinical informatics

[04:06] Importance of workflow in health IT

[05:03] Implementation science

[06:12] Intro to Ambient AI in clinical settings

[10:31] Patient and doctor comfort with technology

[19:15] Balancing AI

[22:32] Rapid evolution of AI tools

[38:53] Shreya’s favorite well-designed thing

[40:32] Outros

 

Transcript:

Dr. Craig Joseph: Doctor Shreya Shah. Welcome to the podcast. Where do we find you today?

Dr. Shreya Shah: Hi Craig. Thanks so much for having me. I am in my home office in the Bay Area in California.

Dr. Craig Joseph: Awesome. And how's the weather? Is it always good in Northern California?

Dr. Shreya Shah: It's pretty good. Can't complain. Nice weather today. We've had some rain, which I think California needs. Starting to feel a lot more like spring now. Just great.

Dr. Craig Joseph: Awesome. Well, it should just be spring all the time. But then you'd be living in San Diego. All right. Well, so you are at Stanford, and you came to my attention, at least, as you put out a paper. And Jamie and I thought that was an interesting paper, among other research that that you've done. Before we get into that, though, can you tell us a little bit about your background? How did you come to do some of the things that you're doing now? You've got a varied career from research to practice. How did you get here?

Dr. Shreya Shah: Yeah, absolutely. So, maybe I can kind of start with my path into clinical informatics, and then I can also talk a little bit about the AI research role that I have. So, for me, my journey with clinical informatics really started through a lens, process improvement.

After starting my first job out of residency, I collaborated on this project to actually update the order entry workflows and our electronic health record for our primary care department. I had no idea that I was actually working on a, quote, operational informatics project at the time. I really just was like, oh, I want to kind of make order entry easier and faster and I think I see some areas of improvement. And that project ended up being so fun and meaningful, and I just found such amazing mentors in clinical informatics.

And so that world opened to me. And then I took on a leadership role as a medical informatics director for primary care and population health, and our chief medical information officer's office ended up becoming board certified in clinical informatics through the practice pathway. And then as I was working to improve those workflows and kind of help with the day-to-day pain points, I also started to become more interested in artificial intelligence and just recognizing these growing opportunities for AI to help solve problems in healthcare.

And that led me to join the AI research team a few years ago called the Heart Team. And they have expertise in AI evaluations with implementation science research methods. And so now we practice clinical care and primary care, do work in operational informatics, and then also the AI evaluation piece.

Dr. Craig Joseph: So, you know, your career path is similar to a lot of folks, at this stage, which is kind of, it's accidental. You were working on a project, and you said you thought it was one thing, but it turned out once you were finished, you're like, oh, I guess I just did that thing. A lot of folks don't really understand the importance of operations, and not only in every aspect of business, but in healthcare specifically. When you say operational informatics, how do you kind of think about that differently than clinical informatics?

Dr. Shreya Shah: Yeah, absolutely. I think they're probably the same, when I think about my hat with operational informatics or I think clinical informatics is just another way to dub that and really trying to help with the clinical operations of the day-to-day workflows. And I almost think of sometimes my colleagues as my clinicians, as almost like patients when it comes to clinical informatics, because really what we're trying to do is try to help solve the problems that they have with their workflow when it comes to their health IT workflows and the way that they interact with, often it is the EHR, and then trying to just help optimize that, make that better. And then of course, the end goal is also trying to make care delivery better for our patients. So, thinking about just all of the complexities, and it's complex but super fun.

Dr. Craig Joseph: It's not always fun, I suspect. But it's often fun, you know, kind of reminds me listening to your talk. When I think of operations, I think of workflow in that you can design the greatest project, but if you don't get the information out to the masses, and if you're not able to kind of communicate the benefits to your target audience, all of that is for not. Let's talk about implementation science. You had mentioned implementation science. I think probably a good chunk of our audience doesn't know there is such a thing. When I was working for an HR vendor, we just implemented, we did our best with our customers. But now there's a whole kind of wing of study. What do you know about that? Are you involved? When did you get involved? Tell us everything. Tell us everything.

Dr. Shreya Shah: Sure. So, the research team that I'm on, they’re really led by clinicians first. I think we think of ourselves as like clinician researchers, clinician educators. And so, we're trained in primary care. And a lot of us have a background in things like UI. And then of course my background in informatics, I think the idea is to really think about implementation as, like you said, bridging those workflows and making sure that when we're implementing any new tool, that we have a good understanding of what is that underlying workflow that we're going to be changing.

Thinking through workflow redesign and then trying to evaluate that and make sure that that implementation is hopefully, you know, a success, that it's not creating any unanticipated consequences. And so essentially the kind of overall goal of what we try to do with those types of evaluations and that our team specifically focuses on AI evaluations. And so, of course, there's lots of implementation science that you can think about with any sort of intervention and healthcare.

Dr. Craig Joseph: All right. Well, you drop those letters A and I so let's get to it. I kind of teased an article at the beginning of our conversation, in JAMA from, maybe 3 or 4 months ago, where you all were among the first groups talking about ambient listening, having an AI write a first draft of a progress note, or at least parts of the progress mode. Tell us everything. What is this ambient AI thing? So how did how did the study come about and what did you learn?

Dr. Shreya Shah: Yeah, absolutely. I could maybe start with sharing a bit about how our team came together. We have this really cross-disciplinary team that kind of intersects operations, intersection research of it, amongst many other groups. And so really the start of our team was actually back in 2023, after generative AI technology started to become more rapidly available, we were one of the first organizations to pilot a GPT four-based tool embedded and the EHR that could draft responses to patient messages that clinicians could review and edit.

And since we were one of the first to pilot it, it was really an opportunity to study how generative AI could transform real world clinical workflows for the first time. And we were interested in understanding the potential impact of the tool. I had lots of questions, what actually helped with the problem we were trying to solve, which was reducing that and basket burden on our clinicians.

And so, through that project, quickly came together. This amazing collaboration across IT, our CMO office, EHR research team, among several other collaborators. And then that's kind of how my two roles actually intersected. And then after that implementation, we continued to collaborate on other implementations. And so ambient AI being one of those and now we're also working on an evaluation of an implementation for another in-market tool that helps draft result comment notes for patient test results in the year. So, we've kind of formed this new team. And I think the coming together was a key part in that partnership.

Dr. Craig Joseph: So, kind of walk us through, if I'm a physician in when you were doing your study, if I'm a physician, what happens. If the patient walks into the room and is sitting in the exam room, what's the story? How does ambient AI work in the real world?

Dr. Shreya Shah: So, it's really impacting the day-to-day practice. It's actually pretty subtle. And I think that's good. We don't want to create too many extra steps for our clinicians.We have our EHR, it has an app on the phone. And then the ambient listening tool is actually integrated into that. Before I walk into a patient's room, I kind of have their chart pulled up on my phone. And then I walk into the room and usually, obviously say hi, kind of check in or introduce myself if I've never met them before. And then I'll actually show them my phone and I'll be like, oh, hey, before we get started today, I just wanted to show you this secure recording tool that we're using to help us doctor notes and see if they have any questions, and then literally just press a button starts to securely record in the background.

So, all that changes is that I'm typing a lot less, like I'm not typing the stuff that all the little things that I'm asking about and the patient's history not typing that anymore. And then at the end of the visit, I just stop the recording and then usually within a couple minutes, once I'm working on my notes, then there will be this draft. I don't think it's generated right into my notes template. It just populates directly in the electronic health record, which I can then edit, and usually requires editing. And then you can go ahead and finish your notes. So that's kind of how it is from that end-to-end workflow standpoint.

Dr. Craig Joseph: So, I noted earlier you practice at Stanford. So, your patients have heard of technology. Do you get pushback from any patients who are like, no, I'm uncomfortable with this?

Dr. Shreya Shah: You know, it's so interesting. I haven't for the most part. When I start describing the tool to patients, they just start nodding like, oh, yeah, in fact, almost a lot of them will be like, oh yeah, I'm using something like that, you know, at work. And then I'll emphasize like, oh yeah, well, this is a secure recording tool. Look, it's like part of your chart. And so that has been a nice aspect. And then of course, hopefully we need to learn more and try to ask patients and get those gather those insights from patients when we're not in the room with them, because, of course, they might have other opinions to share with if we're not actually in that room.

Dr. Craig Joseph: And I like how you kind of frame it, you've used the word secure in every sense. Hey, this is not being recorded on my phone. It's actually going to a secure space, and awesome. So, it's generally well received. And I think that's what we're hearing from people. So, what did your study find? Did it work? People like it. Did physicians adapt and adopt?

Dr. Shreya Shah: Well, so when we designed our study, we were focusing on metrics linked directly to the problem we were trying to solve, which was trying to reduce documentation burden, trying to help reduce clinician burnout. Our team actually partnered with Doctor Shana, our chief wellness officer, and also a national expert on physician wellness. It was amazing to just have him right down the hall to be able to help shape our approach to measuring those things like burden and clinician burnout.

And our results were encouraging in just three months. This was just a kind of quick three-month pilot. We had physicians report a 25-point reduction in documentation, burden on a 100-point scale, and then a two-point reduction in burnout on a ten-point scale. And so those were just really encouraging.

And then beyond the numbers, we also learned through physician interviews and personal stories about how the AI was changing their experience through documentation and just got some great insights about what could have driven some of those point reductions or some of the more fun, kind of why behind those numbers?

Dr. Craig Joseph: It's fun to kind of talk about the successes. Were there any losses? What were any, what are some of the problems? Has anyone kind of pointed out, like, oh, I don't like it. I'm not going to use it because of this or I might not use it for these cases.

Dr. Shreya Shah: What we're seeing is that if among the clinicians that are using it and kind of like using it and start using it, they're actually using it now for a majority of their encounters, often times 100% of their encounters. And that could be both video visit clinic visit.

But then to your point, there's the humans not using it, right? We want to kind of dig in a little bit more and try and understand those groups. A big chunk of who's not using AI right now, we're just clinicians that I've never started to use it. So, for the group that haven't started it yet, one barrier that we found is that they actually wanted more hands-on support during the initial setup process.

To help address that, we ended up partnering with our Epic concierge program, which is, there's a lot of organizations have these types of programs. They offer personalized one-on-one training for ambulatory clinicians for the year. And that program has actually been around for over five years. Our clinicians love it. We have this awesome team of informatics educators that provide that one-on-one support. We ended up embedding ambient AI scribe support into that existing structure.

Clinicians could sign up for ad hoc one-on-one sessions. And then we found that actually helped to just get over some of that initial adoption barrier, help more clinicians successfully adopt the tool.

And then we do have a small group of clinicians that have tried using it. And then are no longer using it. And they've decided the tool isn't a good fit for them. And a lot of times that ends up being no editing requirements. I spoke with a colleague of mine recently in clinic, and was just curious because he was in that group and I was just curious to kind of learn a little bit more.

And he said he found the notes were longer than his preferred style, so he felt like he had to edit. And he's like, I'm efficient with my notes to begin with, and so, it’s not for me right now, but he was open to revisiting the tool in the future. And I think that's actually key, because we know that these types of technologies are rapidly evolving. And so, there's going to be hopefully opportunities to re-socialize with the tool. If someone didn't feel like it was a good fit for them. Once there's updated functionalities and maybe even deeper workflow integration.

Dr. Craig Joseph: The one pushback I've heard that kind of resonates with me, not so much as a pediatrician, but from an internist standpoint, it was with very complicated patients that some docs have said, you know, writing my note, it helps me think and just reading a note that was made by an AI, even if it's a great note, I've lost out on that friction that was built into the whole workflow. And they're worried about that given that lack of the need, you know, if there's maybe a test or an approach that they wouldn't have come to very immediately, but if they have to take ten minutes to write their note, it will come. Did you get any of that or have you felt that?

Dr. Shreya Shah: I think that's so important. And we've thought about that actually more recently, because we are in the process of doing an early pilot with our first cohort of trainees and trying to kind of think about, you know, how do we think about ambient with trainees and medical education, and also what is the potential impact on things, like the things that you mentioned, like clinical reasoning, formulation of a differential diagnosis and assessment and plan, which are all things that are very near and dear to me as an internist.

And I'll admit, I'm guilty of at least in training and in school, having these like, really, long notes as especially with the assessment and plan, just having it be super, super comprehensive. And I think those helped me, I think kind of learn and really learn about those differentials and go to the evidence and try to put together kind of that comprehensive, clinical reasoning formulation.

So, I do think it's really important that we don't lose sight of the impact on those things. That said, at least now what I have noticed once I started my job, I was like, I don't think anyone's reading these notes but me. So, like started to rethink, you know, why do I have these really, really long notes? Like, is it time to kind of rethink my workflow? So, this is predates AMP. But I actually started to shift my workflow.

And of course, you preach a lot of times before a patient is coming in, you sometimes are starting to put together an idea based on why they're coming in and their risk factors. But then I've often started to share my assessment and differential out loud with my patients, like during the visit. And so especially if they're coming in for a new complaint, I'll kind of walk through like, well, here are the different things that could be causing your chest pain, or here are the different things. And here's what I think is most likely. Here are the things that are less likely. But you know, we've got to talk about them because they're more dangerous.

Then I'll type out my instructions and then review them with the patient. And probably a rare person that still prints my instructions out. And no one's taken my printer yet from me. So, I'm glad that I could still do that. But then once ambient was a thing, I was kind of like, okay, this is nice, because in those visits, when I would do that, then ambient would kind of capture all those things. Whereas before, when I was talking out loud and doing all that, I would then go back and then type my assessment and plan and kind of re-rehash all of that after that visit. So maybe there's ways to rethink some of the workflows so we could do a little bit more with patients during the visit.

But that said, I think that those foundational years were crucial for me to go back, take all the data, really synthesize that material and have the time to do that. And sometimes even now, if it's a really complicated patient, you have to go back, gather more data. And so maybe in cases like that, it's going to be changing the workflow to be dictating the assessment and plan. So maybe at the end of the day then there's less typing, maybe less carpal tunnel, but you're still having to kind of actually dictate the assessment and plan for those types of patients.

Dr. Craig Joseph: Yeah. That seems like a prudent approach, which is it's not 100% of anything. It doesn't necessarily have to be all ambient to the exclusion of you being able to actually write or, and when you edit those notes, of course, that that notes presented to you as a draft and you're able to manipulate it in any way. Are you getting more comfortable or maybe you're well past this now, so with respect to your physical exam, are you saying aloud what your findings are or. No, that's not what you're doing?

Dr. Shreya Shah: I'm still not. So, for me, the physical exam as far as ambient goes, I don't think I've kind of leveraged it as much, just because I'm not yet in the habit of kind of saying things out loud, every once in a while, more just to describe it for the patient. I'll say it out loud, but not yet.

Dr. Craig Joseph: Yeah, and I think it's a scribing thing. I remember when I was a medical student and I remember seeing a dermatologist who absolutely was, he had a scribe, a human back in the day who came in with him. And, he said, you know, he would measure lesions and describe lesions using very scientific language and the scribe was recording all of that so that he didn't have to do that.

And it seems to me like I would still find that quite awkward, under a lot of circumstances. But I think people have been doing it for decades. I think it's completely doable under most circumstances. But you're right. Yeah, it's you got to get comfortable with it.

Dr. Shreya Shah: And I think the tools themselves are evolving a bit. So now there's, I think, like we just recently turned on some customization or personalization features where you can kind of make the draft a little bit more customized to your preference. You can now take those exams and kind of pre-template those a little bit, which is what we're traditionally used to anyways, our pre-templated exams.

But then to your point on the derm stuff, I'm thinking now most of the time I'm often taking photos, like taking photos with my phone with the app and then putting those in the chart to maybe in the future there's an opportunity for some sort of, you know, multimodal AI tool that could just help summarize, like doesn't even have to give the differential. But it could at least start by summarizing.

Dr. Craig Joseph: You mentioned the ability of AI evolves so rapidly. Like I don't think we've ever seen things evolve and improve so quickly. I was talking to someone at your EHR vendor a couple months ago and saying I heard a complaint from a doctor about this thing, and the lead developer said, well, I don't, it was maybe about six months ago, and he kind of brushes me off, six months ago.

Oh, my, we can't even talk about something from six months ago. Things change so rapidly. He said, I don't, I can't really talk to you about things from three months ago, because it's not relevant today. And that pace of change, even in technology world, you know, time is unbelievable. And so, it's encouraging that some of the problems that you had. So, you had a physician who said, well, the notes are too long and I'd like it to be tighter and I'm efficient or I'm sure there are. Some are like, well, I like more bulleted problems or I like kind of story-telling as opposed to just the facts. And, and people are able to, sounds like quickly, are being able to kind of tell the AI this and by programing it and by programing it, I mean giving it a prompt. Right.

Dr. Shreya Shah: And right now, it's more kind of binary options. You can make a note more concise. You can make it longer. I do hope in the future it'll take that personalization even to those next steps. But I think one thing that is promising is the fact that personalization, as one of those feed points of feedback, has come up in those early pilots, and that now there's already some solutions to kind of help address that feedback. I feel like that's also really exciting because it means that we can gather feedback, share that feedback, and then hopefully actually see that feedback implemented the solution.

So, I'm hopeful it's going to continue to get better in the future. Hopefully it can even learn from our specific style of notes so that we're not having to select from different options. But also to your point, the bullet points versus having things more in paragraphs like those are also some of the current options that that are available today.

Dr. Craig Joseph: It is moving so quickly. You had mentioned earlier that you're working now on responding to patient messages. And so, just to kind of recap for our listeners, it used to be in the olden days, you called your doctor's office, you might say, I'm calling my doctor, but you were not. You were mostly just calling your doctor's office, and someone relayed a message to them if they couldn't handle it themselves. And with the advent of patient portals, a lot of physicians started seeing a lot of messages that typically they wouldn't have seen. It would have been handled by someone on their team, but it was easy just to send it to the physician.

Thankfully we're improving that now, and a lot of those messages are being handled by others before they get to the physician. But for the ones that do make it to the physician, you are now able to, under certain circumstances, have a draft response before the doctor even looks at it. How is that working out?

Dr. Shreya Shah: Yeah. So that tool we've had turned on now since 2023 and, started in a couple specialties, and then we're in the process of expanding that one. I think it also had some of the similar outcome metrics, like showing reductions in cognitive burden and also improvement in burnout scores like we saw with ambient AI. So, it seems to be a win.

So, I think that one we're just in the process of trying to kind of scale and expand that, and then also just looking for other areas of opportunity, other kind of similar problems to solve, recognizing that the end market is such an area of burden. So that's where we've been doing this recent pilot with this similar type of tools are very similar to the patient messages, but in this case that helps to draft a summary of the results for patients for their test results.

And so, it could be lab or imaging tests. And right now, you as a clinician, you get those in your market in a separate folder. And then you can review those and then put a comment, to explain the results to your patients. And so now we're also integrating this option to have each test result. That could be draft test results that could be shared with patients.

Dr. Craig Joseph: That's hard. Yeah. It's much easier to summarize an office visit, but lab results is as, a normal result is not necessarily if it's in the normal range. Doesn't necessarily mean a good thing. Right? It often does. It almost always does. But there are patients where like, no, no, no, no, we need that number much lower than the average person because you have this specific disease. And so that takes a lot more clinical insight. And often I think it's probably not that the lamb or the I can't do it but it to be successful it would need to have a large amount of information. And I think what I've heard, one of the restrictions has been in the past that fewer numbers of tokens. So, it was difficult to give the AI access to all the information that could help but give it more context. Where are we now with that?

Dr. Shreya Shah: Yeah. So, for this tool specifically, it's actually a tool that was built in-house by our AI team and showed that given us this unique advantage, because we can customize the tool from the ground up based on clinician feedback. And so, before we did this larger pilot that we're doing now, we did some kind of clinician co-development with clinicians and then with the AI team where we had a little over 30 clinicians use the tool, and then we'd have frequent feedback checkpoints.

It was really fun to see the iterative process because it was developed in-house. It changes from the feedback through like a change process. Control could actually be incorporated pretty quickly though, compared to of course, if it's not in-house. To your point, like one nice example of one of the specific points of feedback early on was that the tool was missing important clinical context, like especially if the patient had had that lab test previously or had that imaging.

Because we know trends are a really important thing that we look at when we're interpreting those test results. And so, the team was able to act on that feedback and then kind of update the way that model works. And so now the draft comments incorporate historical data for those trends. It's just one of these nice examples of being able to really get the clinician feedback, but then iterate to try to continue to improve the tool.

Dr. Craig Joseph: Yeah. And that combined with the fact that it's so quickly becoming easier to incorporate more data, that used to be the restriction, right. You can have all of this, but it's going to cost you a gazillion dollars. Well, then that's not realistic. We can't do that. But as those prices come down, and they're predicted to come down. But boy, they're coming down very, very rapidly.

It's easier to give more data. And with more data comes more context. And it's good to have the lab results for the last five years, but also good to see what the physicians wrote in their notes about those lab results. Right. And then you do get to that sweet spot where you're able to actually predict what I would say. Right? That's when you're most of the time predicting what I'm going to say, I trust you. I used to joke I had a medical assistant who had been in pediatrics a lot longer than I had. And this is all in the paper world. But just to kind of give a sense of, I would sometimes walk in a room and again and there would be a paper chart in the door and there would be a patient handout, there would be a handout explaining croup or chicken pox or whatever it was, strep throat.

And I said, well, my medical assistant, mind you, not nurse, medical assistant or a medical assistant, has diagnosed your child with croup. I will now check because she is only right in my experience, about 98% of the time. And so, I need to double check and make sure she's, this is not one of those 2%, times. And, and she got to know me, and she was able to predict what I would say. And she was also able to differentiate what I was going to say versus what my partner was going to say, because we often, every physician approaches things with a slightly different spin. And, when we can start to get technological tools that do that. Well, it's, game on.

Dr. Shreya Shah: No, you're absolutely right. And I think another and based on feedback, another way that they updated the tool was to not try and give like information about next steps or recommendations to keep it objective, because we know that the next steps can of course vary by the clinical context.

With both this tool the results, and with ambient AI, it's still going to come back to trying to make sure that clinicians can still focus on those things that require the highest skill, like clinical reasoning. You see an abnormal test result, and now you go through the algorithm for what to do for that, but maybe helped streamline the more administrative like if it's especially for normal test results, make sure that patients are getting some nice, patient, friendly verbiage that you objectively explained. The result. But absolutely thinking through ways that this is not going to manage results for us, but maybe just assist.

Dr. Craig Joseph: Yeah. And add the color commentary, the flowery language that of so many patients like that, so few physicians have time to address. And, you know, there have been some studies that the people like the messages that they get from an AI, or the responses more than a human. And I think that that's not that humans, the doctors, are incapable of doing it, just that we don't have enough time. And so, it becomes less important.

So, it becomes more of a here's the facts as opposed to a we're going to add some flowery language and really want to tell you that you're quitting smoking is why these lab results are looking better now. It's important to do if you had unlimited time like an AI does. And while we're almost done, let me ask you, a penultimate question here.

You work in Silicon Valley. I'm sure I will encounter some of these vendors that you're using now and others who have ideas about AI and healthcare. Any advice to that person in their garage just down the street from you about how to make sure that they understand workflow, that they understand being human centered and usability, any high-level advice to those folks?

Dr. Shreya Shah: Yeah, absolutely. I think it just comes down to that, trying to think about code development as much as possible and really think about who's going to be using the tools, who are the key stakeholders. So, clinicians, patients, engaging them in making sure that they have, some sort of way to help shape the tool and the design of the tool. Really having a deep understanding of the workflows, like you mentioned at the beginning, because otherwise you can't redesign workflow without really understanding the pain points. And what are those current state workflows. And then probably of course, thinking about what can I help drive someone to want to take that on like that gets to some of those outcome metrics or what's the problems they're trying to solve?

What's the ROI kind of benefit can it be? And that's working with some of the stakeholders and healthcare system leaders or things like that. But I was actually a critic, of course, this past week, and I was thinking of an example where I was like, this could be done much better. And I kind of had a sense of that before they came in, so I was. And it's been a while since I've done a TIA workup, so I wanted to just make sure I am up to date on the stats actually pulled up to date, which is kind of the one. There's the clinical knowledge system that I use going through the articles on TIA workup.

And then when I'm seeing the patient, I'm like talking to them, examining them, confirming that, okay, we should probably go do this to work up putting in each order separately. During that visit, I even pulled up to date again just because I was like, oh wait, there's this one. Let me make sure I'm ordering, you know, the imaging study correctly with the right protocol. And I was just like, there's got to be like this. I'm sure it's something that we can do better because there are these really great pieces of evidence based out there. Ambient has the potential to integrate into our workflows. Ambient knows that I'm talking to this patient and mentioning that I'm worried maybe this was a TIA. Could it pull up some of those algorithms for me in the charts? I'm not clicking into different screens. Could it then just tee up the orders for me? Associate diagnosis code as I'm talking to the patient? Could it just when going through the instructions and be like, all right, you're going to call this number to schedule this. But I just draft those instructions because right now I'm still typing them. So, I feel like just even shadowing a clinician and kind of seeing and the gamba could absolutely go far in, in trying to help solve real problems.

Dr. Craig Joseph: There's a vendor that is working, I think, on pulling in guidelines based on the conversations that you have, that now, that's where it stops. I don't think there's anyone that's queuing up orders or queuing up patient advice station yet.

Dr. Shreya Shah: Hopefully it is.

Dr. Craig Joseph: Oh no. Oh yes. I mean very clearly. Very clearly. Yeah. Well, that's good advice. And one that is well taken, really understanding what the problem is you're solving from everyone, all the different perspectives is essential. Otherwise, you might be solving a problem that doesn't really exist. And you don't typically do well as a company when you're solving a problem that no one really has or is not important.

It's time for our last question. We always must ask the same question to all the folks that we talked to, what is something in your life that's so well designed it brings you joy whenever you use it or interact with it?

Dr. Shreya Shah: As I thought about this recently and was actually trying to really think about, like, what am I actually excited about it? Or what's actually kind of making me smile? So, I ended up thinking about this app that my son's daycare uses. He's two and a half and it's this app called Bright Wheel.

The app is really easy to use from the parent and like, it's, you know, no training required. You kind of have everything that you need in there. They, the teachers use it to update things like nap times and what they ate that day. And they can also send messages to us if they need to update us on something. But my favorite, favorite part about it, and what actually makes me smile, is you just will randomly see a photo come in, or a video come in of them at daycare, and it's just like it actually is the time of the day where sometimes during lunch I'll be like, oh, are there any photos or videos? And, you know, just to be able to pull those up.

And of course, we take so many photos and videos of our son that I'm always dealing with like storage capacity issues on my phone. But I feel like there's something about I didn't take these and I'm not there, and I feel like it just helps me maybe connect with him in a way during like a busy week when, you know, we're at work and just yeah, that's I think something that brings me joy. And it's simple and easy.

Dr. Craig Joseph: That's great. I wonder what you should be able to do is take a picture of you at work and send that.

Well, Doctor Shreya Shah, it was amazing talking to you. And you know, we look forward to you solving all the world's problems or at least some of the eye and physician burden problems.

Dr. Shreya Shah: Thank you so much. Again, so fun talking to you. I learned a lot as well. So, thanks for having me on.