Structural racism in health tech [PODCAST]

Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes

“We focus on ensuring our solution works equally well for English speakers with any accent. We regularly hear from doctors how much of a difference this makes in their ability to deliver high-quality patient care. It’s our hope that more health tech vendors will take a good look at what they’ve chosen to prioritize and how that may contribute to inequities within the system. The factors which inspired the American Medical Association to recognize racism as a public health threat demand that we all play our role in the multi-faceted structures that contribute to delivering equitable health care to all Americans.”

Matthew Ko is co-founder and COO, DeepScribe Inc., and can be reached on Facebook and Twitter @DeepScribeAI.

He shares his story and discusses his KevinMD article, “Examining structural racism in health tech.”

Did you enjoy today’s episode?

Please click here to leave a review for The Podcast by KevinMD. Subscribe on your favorite podcast app to get notified when a new episode comes out!

Do you know someone who might enjoy this episode? Share this episode to anyone who wants to hear health care stories filled with information, insight, and inspiration.

Hosted by Kevin Pho, MD, The Podcast by KevinMD shares the stories of the many who intersect with our health care system but are rarely heard from.

Transcript

Kevin Pho: Hi, and welcome to the show, where we share the stories of the many who intersect with our health care system but are rarely heard from. My name is Kevin Pho, founder and editor of KevinMD. Today on the show, we have Matthew Ko. He is the founder and COO of DeepScribe Inc., and he wrote the KevinMD article “Examining structural racism in health tech.” Matthew, welcome to the show.

Matthew Ko: Thanks for having me, Kevin.

Kevin Pho: We’ll get into the article in a little bit, but first off, can you share your story and journey to where you are today?

Matthew Ko: So I’m one of the co-founders of a company called DeepScribe, and at DeepScribe, we essentially automate clinical documentation. Our product records a natural patient conversation, and through that natural patient conversation, we’re able to generate a comprehensive patient chart. And the story of how we came to it is pretty interesting.

I like to introduce myself as not a doctor. My background is actually in technology. I spent the majority of my career in consulting for data analytics, and we spent around a year and a half of research and development at the Berkeley AI Research Lab developing the core technology that fuels DeepScribe. And we fell into this problem really from a patient’s perspective.

When I was doing consulting, I was also serving as a care coordinator for my mom when she had breast cancer. I lived in San Francisco, and I was flying to Los Angeles, where I grew up, almost every week just to be with my mom, and mostly to be my mom’s translator. My mom is from Taiwan, and she doesn’t speak English that well. We like to joke that the idea of DeepScribe started when my mom asked me if I was lying to her about her treatment. At first, I thought that maybe my Mandarin was a little bit rusty and she didn’t really understand what I was saying, but, you know, it became pretty clear to me that she was being serious. She would say to me, “Matt, you’re telling me everything is going well, but I’m not getting the same nonverbal feedback in the room.”

And it wasn’t until my mom’s oncologist answered a text message in the middle of delivering her prognosis that I just became an evil patient. I’m sure you and some of the listeners know evil patients. They go to Google to get their MDs, and they would walk in and try to tell the doctor what they’re doing wrong. I became that patient, and it got to the point where I wanted to transfer my mom’s care to somebody else. That’s when I went to my co-founder’s father, who was an oncologist, to transfer my mom’s care to somebody else. And it was really through him that I was able to appreciate the problem that we’re solving today, which is clinician burnout, and really understand what these physicians need to go through on a day-to-day basis just to be able to provide high-quality, compassionate care to their patients.

Kevin Pho: Speaking from the perspective of health tech, what are some of the biggest misconceptions that you want to convey to my clinician audience about the health tech side of health care?

Matthew Ko: The first thing that I like to say is, you know, technology is an investment, right? I think a lot of physicians that I meet expect technology to be almost this miracle, where the moment you start using it, it’s going to work for you. And one of the things that I notice is that, you know, clinical workflows are sacred, right? For a physician to change something that they’re so accustomed to, there are a lot of growing pains involved. And a lot of times, I see physicians just kind of check the technology on the side of the road because they’re not willing to make that investment.

However, the physicians that I do see make that investment are able to really ameliorate their workflow. So, you know, our physicians tell us that for the first time ever, they do not have to choose between compromising their own well-being and providing high-quality, compassionate care. And I see that with a lot of other technologies in the space, where my friends are founders. Getting the physicians closer to the left side of the innovation adoption curve, rather than the right side, is the hardest challenge in technology in medicine. And that’s why you see technology being almost, you know, the least pervasive of any industry that we live in.

Kevin Pho: So, speaking from the physician standpoint, I think a lot of the reasons that explain what you say are that sometimes, when we make that investment in terms of changing that workflow, the payoff isn’t necessarily better than the status quo. So I guess, when confronting new technologies that require that investment, how can physicians be sure that that investment in time, and that disruption to workflow, will in fact be worth it at the end?

Matthew Ko: I think a lot of that comes from, number one, the numerous unmet promises that physicians have dealt with. I think, starting with the advent of EHRs, right, physicians were promised all these lofty things. And when the digitization of medicine really became the status quo, it actually led to the physician needing to do a lot more things that they previously did not need to. And I think that this taste really carries on to a lot of the more innovative technologies that we see in medicine today.

And I think, number one, knowing that the technology where it is today, at least natural language processing, at least in artificial intelligence, deep learning, you know, is truly something different than it was back then, where people were, you know, almost marketing this AI theater, right? I think now technology has really submitted a peak to be able to transform health care.

But to answer your question, I think the first thing is just social proof, right? I think one of the things that really helps our physicians get over the hump is being able to talk to other people that have made that investment. When we give references, we try to find people that are, if we can, exactly in the same town, or at least in the same specialty, on the same EHR, with a similar workflow. I think starting with the social proof and asking for that is a good first step, and making sure that that company is the right partner for you. It’s not just a matter of picking a technology that is shiny and looks good on paper, but really talking to the people that are behind the company and seeing if they hold the same philosophical values as you.

Kevin Pho: All right, let’s transition into the KevinMD article that you wrote. It’s titled “Examining structural racism in health tech.” Now, for those who didn’t get a chance to read that article, can you just walk my audience through it and share the story of why you decided to write it?

Matthew Ko: Sure. Yeah, so the article was really a response to a tweet that I read by the editor-in-chief of JAMA that said no physician is racist. And, you know, I felt the need to respond to this, as many others did, but I thought that I had a unique perspective as a technologist that was worth sharing. I think a lot of times, physicians talk about, you know, implicit bias with patient care, but I think the topic of algorithmic bias in health care is not really discussed, right? And I think that the algorithmic bias in medicine has really trickled down to the fabric of our health care system and has really dictated the way certain physicians practice medicine.

So, you know, an example of this that I discussed in the article is, when creating DeepScribe, we interviewed hundreds and hundreds of doctors when we built our product, and we realized that, like you said, there are a lot of doctors that tried these new, innovative technologies, and they failed them, right? And a lot of these doctors tend to be the ones that are kind of marginalized from where you see the majority of the data that is used to build these technologies.

A good example is voice recognition. Today, voice recognition is largely seen as commoditized, but when it was first released to physicians and marketed to physicians, it was almost seen as this miracle, right? And a lot of these physicians that we interviewed, you know, are not native English speakers, and a lot of these dictation tools failed them. As a result, what ended up happening was they said, “I’m never going to,” you know, like you said, “invest in this technology again.” And it really dictated their workflow from then on.

So what you see is that the way a lot of people practice medicine is ruled by the amount of technology that they have available and how well that technology works for them. And I think it trickles down to also how providers practice care, right? For a physician that is very savvy on dictation today, because they’ve been using it for, you know, 10, 15 years, they’re able to give some more attention to their patients. And I talked to a lot of doctors who, before, were just staring at a computer, and they would tell me, “Matt, I did not realize how much I was missing by just staring at the computer. After using DeepScribe, I’m always looking at the patient, and I didn’t realize how much I was missing from providing care just by, you know, not looking at the patient.” And I think the algorithmic bias has led to some of the implicit bias that physicians are facing today.

Kevin Pho: So tell me what the current scenario is when it comes to health tech startups. Are they aware of this issue of implicit bias? And whenever they do survey clinicians, do they take a cross-section across socioeconomic lines? So what’s the current situation like now?

Matthew Ko: I think it’s often overlooked, Kevin. And, you know, that’s one of the calls to action that I had in the article, which was, as technologists, we also have a social responsibility to think about a lot of these things and think about the impact that the technology we bring to the market makes. One of the biggest surprises that we had was that we unintentionally, not solved the problem, but we found that our technology unintentionally worked for a wider breadth of physicians that come from, you know, non-English-speaking backgrounds, because the way that the technology works did not rely on the precision of the transcription, but rather on the ability of our algorithm to identify the relevant entities inside the transcription.

So, you know, I don’t think that the startups are thinking about it today. However, I do think that the level of sophistication, at least in voice recognition, has improved. If you look at some of the modern-day technologies, they are almost synthesizing multiple neural networks to be able to cater to different dialects of English. So Dragon today has vastly improved, because the way that it works today is it would actually detect what type of accent is being said and merge them in real time, rather than saying, “Hey, I need to buy this specific, you know, Mandarin English version of Dragon to use.” It’s combining all of their models at the same time, and that leads to better performance.

Kevin Pho: We’re talking to Matthew Ko. He is co-founder and COO of DeepScribe Inc. He wrote the KevinMD article “Examining structural racism in health tech.” So Matthew, looking forward, what specifically do health tech companies need to do in order to avoid the structural racism or the algorithmic bias that you talk about? And how are those solutions going to disrupt your workflow in terms of that path to innovation?

Matthew Ko: Yeah, I think the first thing is just really being aware of it, right? So just thinking about it. And I think, you know, bias in AI is an up-and-coming topic right now that a lot of people are thinking about. I think that actually solving the problem is a little bit more difficult, because you have to be a little bit more careful about the data sets that you’re using. So, you know, especially in voice recognition, if they’re learning on all English-speaking voice samples, it’s going to perform better in English voice, and it’s going to actually perform worse in non-English, of course.

So I think the first thing is to make sure that one of the features of your data set is to look at whether or not you have a population that represents the entire population. And that’s one of the first things that we’ve had to do. For us, it actually came naturally, because, you know, I myself am not a native English speaker, and my co-founders aren’t either. So the first guinea pigs for our product were actually all non-native speakers, and our first, you know, kind of core data set was actually majority non-English, and it later became more English.

Kevin Pho: All right, and my final question: What’s your take-home message that you want to leave with the KevinMD audience?

Matthew Ko: The care that the patient perceives is the care that the patient receives. That’s one of the things that I learned, you know, serving as a care coordinator for my mom. I know that a lot of times physicians are burned out and they’re just trying to get through the day, and a lot of times these physicians need to compromise their own well-being so that the patients can perceive a better level of care. However, at the end of the day, I think it makes such a tremendous difference to the patients. The studies have shown that the patient’s perspective of their own present illness really dictates a lot of the outcomes, and for my mom, I know that was true.

So, you know, I think, although it’s difficult, I would say look at technology as an investment to help you achieve that. And hopefully one day we’ll see technology as pervasive in health and health care as we do in the other industries.

Kevin Pho: Thank you so much for sharing your time and insight, and thanks again for being on the show.

Matthew Ko: Thanks, Kevin.

Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes