How to Transform Monitoring and Diagnostics with Voice-Based Technologies without Violating Privacy with Jim Harper, Co-founder and COO of Sonde Health
: [00:00:01] Welcome to the Outcomes Rocket cast where we inspire collaborative thinking improved outcomes and business success with today's most successful and inspiring healthcare leaders and influencers. And now your host Saul Marquez
Saul Marquez: [00:00:19] Outcomes Rocket listeners, welcome back once again to the outcomes rocket podcast where we chat with today's most inspiring and successful healthcare leaders. If you like what you listen to today or you like what you listen to in general with the podcast, go to iTunes podcasts and give us a rating and a review and subscribe. We'd love to hear your feedback and always are looking for ways to do things better or keep doing things that you're enjoying on the show. So without further ado, I have an outstanding guest for you. His name is Jim Harper is the co-founder and chief operating officer at Sonde health. And so Jim has just a wealth of experience in this industry as an entrepreneur he's had an experience here at San health for the last almost three years but he's also been an entrepreneur in residence and also has a deep understanding of bioengineering systems. With his experience at MIT. So before I go into it any further I just want to give you a warm welcome Jim and open up the microphone to you.
Jim Harper: [00:01:22] Well thanks, Saul. And I think it's great to be a part of the conversation you've started on outcomes rocket and you know just a little bit of context for my background and why. As somebody has formally trained as a biochemist and with a long background in technology development that I'm so interested in health outcomes it's really started back with when I was studying neuroscience and looking at Alzheimer's disease and mechanisms. I realized that for me basic science and the big questions there had some limits and there were other big questions that I wanted to be able to answer. But we just couldn't we didn't have the information tools. So I've been really excited about both the drive to improve healthcare as well as the convergence of technologies that may help us do it. And so that's why I'm here.
Saul Marquez: [00:02:07] That's pretty awesome. And the meandering road of research and just biochemistry brought you here just to create a bridge to really fulfill that missing gap that you experienced.
Saul Marquez: [00:02:18] And Jim tell us a little bit about sonde health and what you guys do there.
Jim Harper: [00:02:22] Yeah so sonde health is a digital medicine company where we're developing voice based technology that has the potential to transform the way that mental and physical health is diagnosed and monitored. And I think our goal is really to be innovating at the intersection of voice interactive devices machine intelligence and AI systems and the overall healthcare system to enable what we call persistent health awareness. The technologies that are always sensitive and always secure. So we've been demonstrating that proprietary analytics that are running on devices that are already in almost every pocket and entering more homes every day have the potential to transform what are often imperceptible changes in the sound of our voice. So how we say something not what we are saying and transform those into objective and quantitative information about a range of important health states that affect the nervous muscular and respiratory systems and maybe even more.
Saul Marquez: [00:03:17] And so are you thinking like tonality things like that.
Jim Harper: [00:03:20] Yeah. So their speeches are a really interesting thing for a variety of reasons. One it's unique to humans and it's really the most complex thing we do it involves more active muscle groups operating at the same time than any other physical action that we do it involves large numbers of brain circuits kind of proportionally for actual areas of the brain to produce effective speech. And then it requires a healthy respiratory system to activate the source. And so what we're looking at are how changes in the physiology of those three major systems are reflected in acoustic changes. So back to your earlier point it can be changed in pitch slow changes in intensity changes in pacing of the speech and even more so. You know we can describe mathematically thousands of different low-level features that represent different aspects of those acoustics and only a subset of those are we seeing in our investigations very when the physiology of those systems change. And so finding those and appropriately weighting and combining those with the best analytical methods are we able to then get outputs from this voice analysis that correlate very well with existing best measures in health.
Saul Marquez: [00:04:37] That is super interesting and just to think that the technology necessary to do this is already in many of our pockets.
Jim Harper: [00:04:47] That's right. So you know I think the convergence that's happening we're all very aware and we've seen more of it every year. The computing power in our pockets exceeds what took us to the moon. And I think that just within the past year the explosion of voice services from Apple Amazon Samsung.
Jim Harper: [00:05:06] You know a range of people has shown that the underlying capability to analyze voice in real time exists at least with respect to analyzing the content of we say now we can argue about how you and what we're able to do with that so far. But I think the trajectory is universally seen as very positive and I think what we're trying to do is offer a new dimension of information beyond just the linguistic content that can help move these technologies from just menial assistance that automate things that we could already do to ourselves to provide more transformational capabilities that disrupt things like healthcare.
Saul Marquez: [00:05:43] This is interesting and Tommy. Jim what do you think around this topic should be you know just something that healthcare leaders should be focused on.
Jim Harper: [00:05:52] So being newer to healthcare right longtime and technology I've been talking to a lot of leaders in the space and going to a lot of places where they gather and I hear conversations about transitions from volume value-based care and moving from reactive to preventative patient management philosophies.
Jim Harper: [00:06:12] Yes but within that you know as I look at it with my system perspective what I see is a glaring lack of data on what is happening in our health day to day with a frequency that is sufficient to help us understand what does the transition from health to the disease look like and how can we measure it in a paradigm that isn't what we think of today as diagnostics. And so I think translating that into a different question is if we're going to seek better outcomes and in ways that are measurable what is it that's going to allow that measurement to be affordable and to understand how to quantify whether prevention is working.
Saul Marquez: [00:06:54] Yeah. And so the thing that comes to mind Jim is we've got these awesome technologies. How do we scale them? How do we make them available? What are your thoughts on that?
Jim Harper: [00:07:03] Yeah. And so for me right it started a number of years ago I was excited as everybody else thinking about the potential of wearables right to really address the gap in information and after wearing between 5 and 10 wearables at a time for a couple of years in a couple of conclusions and one is even for somebody as motivated as I am to understand the burden of operating what are otherwise simple devices is actually quite high and two the cost even if they are 100 dollars a person if you scale that to population scale health management it increases the cost of that at a time where it's really kind of addressing the unsustainable increase in the cost.
Jim Harper: [00:07:45] That is the driving motivation. So I just found those to be very inconsistent. And what I saw as the white space and I think pure tech you know when I joined them or was seeing the same opportunity. Is how do we eliminate the cost and the device burden and the user burden with collecting this information by enabling new analytical technologies on devices. People already know and so you look at the space and people have tried and I think successfully in some domains to utilize accelerometers or activity on the phone to get insights into behavior which is appropriate to behavioral health in some ways but to be really relevant we need what I think is the equivalent of the CBC or the blood panel you know how can we take single samples or small numbers of samples from people that don't require them to change their daily activities but derive a number of meaningful tests that give us the equivalent of a thermometer. Right. We don't have to have a true diagnostic but we need to understand better how to allocate scarce resources and costly resources to the patients that will actually benefit from that. And I think that's a sign the driving force behind why we make choices on the way to implement the technology. And I think the value propositions that we see most exciting emerging from it.
Saul Marquez: [00:09:01] Now the really interesting Jim and you bring up a good point right because I forget what the statistic is but the majority of healthcare is light comes from 5 percent of those in the system.
Saul Marquez: [00:09:11] And if you know if we could focus the devices on that small niche population that's causing most of the expense.
Saul Marquez: [00:09:20] Maybe that's where we start.
Jim Harper: [00:09:21] Now and I think you're right. So for things like wearables that I was talking about earlier where there is some cost into implementing the systems. That's absolutely right. There are best addressed to the sickest people in the system.
Jim Harper: [00:09:31] If what you're trying to do is not prevent healthcare transactions and not prevent costly care episodes which I think is implied by targeting those sickest individuals get us what we want to do is improve outcomes by helping people not to need that care in the first place or not to transition to disease. And I think the question is different and I think that's where the economy really changes and where passive technologies that essentially are free to implement derive their value from the utilization of the information and the services that they cue is Where I think I'm excited about the potential. It's a longer term potential to be sure. Yes, it's one where the time is now from a technology perspective. And I think that progressive leaders in healthcare who are really looking to create new models of care that are based on preventing disease are the kinds of partners that I think are ideal for this technology.
Saul Marquez: [00:10:26] It's really fascinating to hear a perspective on that Jim because yes, on the one hand, my approach with my suggestion was yes you know let's go to that population. Let's cross.
Saul Marquez: [00:10:36] But you're saying let's keep people out of the health care system let's keep them healthy through the use of technology.
Jim Harper: [00:10:42] Absolutely as a long term goal I think that's where we all want to be in terms of like the ultimate outcome is to prevent us from having to encounter the system and those costly ways.
Jim Harper: [00:10:51] Now in the road from here to there I think that there are several addressable places where we can begin that process and looking at kind of conditions that are traditionally difficult to diagnose that is heterogeneous in the kind of way that the disease manifests in individuals and have long lags between onset of disease and diagnosis.
Jim Harper: [00:11:14] Those are really interesting places to start using the technology that bridge kind of near term value propositions but also demonstrating the long-term potential of constant persistent health awareness to address needs. So a few examples. But I tend to think about. There are in neuroscience where a lot of the assessment and diagnosis today is heavily reliant on interactions in the clinical setting not just on diagnostic tests. So for Alzheimer's disease, Parkinson's disease, depression, those fall into this category. But it's stunning when you look at depression the diagnostic lag the time between the first episode of depression and when somebody actually engages the system in care is on average around four years. So when we talk about taking bergan out of the system and improving outcomes if we can do something to reduce that lag and reduce the number of episodes I think we're doing something very meaningful that's addressable a really good example right in Alzheimer's. Those kinds of numbers and diagnostic lag from first onset of cognitive impairment may be on the order of 1 to 1 and a half years in Parkinson's it's more like a year. It's a process of ruling out a lot of other things that take a lot of time and burden. If we have objective measures that can help winnow down and narrow those they don't have to be the diagnostics themselves to have significant value and improving outcomes.
Saul Marquez: [00:12:40] Really really interesting yeah it's a good distinction that you made there. Jim So would you give us an example of what you've done there at Sonde health to apply some of these ideas to do things differently and better outcomes.
Jim Harper: [00:12:54] Yeah so let's talk a little bit about the work we've done in depression so far. So if we look at screening for depression the United States Preventative Services Task Force last year put out two recommendations that suggest that all adults and adolescents in the United States be screened at least once a year for depression in recognition of the burden associated with untreated disease.
Jim Harper: [00:13:16] But if you look at the latest Ambulatory Care Survey the number of individuals who are actually being screened is around four point two percent. So we have you know a 96 percent addressable market there in terms of people who aren't being captured by the existing screen. There are a lot of reasons for that but we feel like and I love your conversations with other people about access to care because access to something like that screen is a big part of that. And one question we have is why does not screen in satisfying that necessarily have to be in the care event or the visit to the primary care physician. Why can't it be done in close coordination with it. So the 9 is the most common instrument that is used it's a 10 question questionnaire and you answer multiple choice depending on the severity of how you've experienced symptoms over the last couple of weeks. You some the score and then over a number of studies they validated threshold score of 10 or more. That indicates moderate to severe risk for depression. So in practice that's the most commonly used screen. What we've been doing is doing studies where we offer our research tool which is an app that's on iOS and Android phones asked people to provide basic voice samples so these can be as simple as repeating ah for five seconds or something like Potapov like a product which is called a diode kinetic pasque. We're just getting voice samples. We may ask them to read a sentence or just give us a prompt in speech.
Jim Harper: [00:14:47] Then we ask them to complete the pH Q9 which gives us the training reference and what we have seen is in thousands of subjects. Now to date who have completed this study with us that we can take as little as 10 seconds of speech and derive the vocal biomarkers create models that predict the pH Q9 score that agrees a high percentage of the time. So now instead of requiring people to recall their symptoms sometimes, that's imperfect or complete the questionnaire. Now we have something that's consistent with a quick voice sample. Yes. They even pass it right. So being able to listen and derive those features. Now I want to be really careful when I introduce that that's not what we're doing today. We're being very careful because we understand. Right. And I have the same concerns. There's a lot of privacy and security concerns around this kind of monitoring. And so it's really important to us that before we try to present that kind of technology that we ensure that it is capable of being always sensing always secure. And there's a number of things that have to happen there in terms of using the right kind of linkage and biometrics to say the individual speaking as somebody who's consented to that measure. So those are misused and also making sure that the accuracy has been validated rigorously scientifically so that any answers we provide are no.
Saul Marquez: [00:16:09] Jim that's really interesting. And then the other idea here is OK so we've got the power of this sensing technology the voice biomarkers maybe we even use the power of the pen and prescribe spot checking for this maybe a couple of times a week.
Jim Harper: [00:16:24] It's possible you know and I think what we envision great and I think really what's possible with the technology has provided that we address the privacy and security concerns adequately is a world where your Siri your Alecks your you know all of these voice interact and substance instead of waking up to a keyword they could actually wake up to a health condition like the onset of depression or the onset of Alzheimer's disease. And cue you encourage you to interact with the healthcare system in appropriate ways so we don't see that as in any way replacing the doctor. But the goal for us and I think a really interesting question as we try to drive better outcomes is how do we change the equation from requiring that patients know when and how to seek care and make it possible for care to effectively seek those patients.
Saul Marquez: [00:17:18] So let me ask you, Jim. Who are you looking to interact with? I mean you've got a big project on your hands. Who are you looking to interact with to make this happen? Because it sounds like a really great idea.
Jim Harper: [00:17:30] So I think around depression screening for example if we are working and having conversations with integrated delivery systems that are trying to integrate behavioral health into primary care in meaningful ways. And I think implementing this kind of system at the beginning first through the pH Q9 but ultimately with an eye toward augmenting and then replacing when validated with the voice system we learn together about how to provide access to patients in meaningful ways to screening. And for us too there's an added kind of motivation that the information and the screening by itself is not necessarily the whole story but it creates the opportunity for new timely feedback loops in care. I think all of us have experienced you know you go you get a prescription you go home. But the follow up and the feedback is often difficult and challenging the system and here there's an opportunity if somebody screens with mild risk of depression cognitive behavioral therapies other things that can increase access to appropriate care and be delivered on the same platform now become really interesting and meaningful opportunities to explore with partners who are trying to improve persistence.
Saul Marquez: [00:18:43] Interesting yes. So to the listeners if you're looking for a new way or are just another partner to help in your efforts in this area. By all means, consider Jim and his team at sonde health to create that part of the reason we created outcomes rocket is to help break down those silos and give you access to people thinking about ways to improve healthcare and improve the collaboration that that happens within this space. Jim thanks for sharing that. Sure. So tell us about an exciting project or focus that you're working on today now.
Jim Harper: [00:19:17] So I think one of the exciting things we are doing is what I was describing it is really moving beyond the cued kind of samples that we're talking about and building in the features necessary to listen.
Jim Harper: [00:19:32] On health we like to say your health speaks and who want to listen. And so the other projects that are exciting are partnerships with healthcare organizations that have a large number of people with a range of diverse conditions and to be able to discover together new vocal biomarkers so who we are working both with pharma companies as well as academic researchers looked at local biomarkers and a range of other conditions.
Jim Harper: [00:20:02] So I mentioned Alzheimer's and Parkinson's disease not by accident or efforts that are beginning there.
Jim Harper: [00:20:09] And I think we the exciting thing to me about this vocal source is that we want people to bring us their problem I think there's a lot of ways for creative data analysis to happen and the burden of implementing the system is so low that there are great opportunities.
Saul Marquez: [00:20:25] I'm just curious and thank you for sharing that Jim about vocal biomarkers. I mean how do you determine that in a nutshell. Like for somebody, it doesn't know the power of that. Maybe you can give us like the 101 here.
Jim Harper: [00:20:37] Yeah. So I do that or use biomarkers intentionally by analogy to what we traditionally look at as follicular biomarkers that we will analyze in blood or saliva or genetic biomarkers of disease risk.
Jim Harper: [00:20:51] I think in very similar ways we are seeing these subtle changes in the voice that we can measure quantitatively objectively have a similar predictive value in terms of the presence of a condition and or potential progression of the disease. So we perform and we look at the science in doing studies in the same way we look at large numbers of population individuals with and without a condition to correlate these specific acoustic features to the presence of that condition. So each one of these low-level features then become independent biomarkers that we can combine to achieve different predictive outputs.
Saul Marquez: [00:21:31] That's fascinating and kudos to you guys for just being able to decipher those specific markers that are indicators of disease or depression. I'm really intrigued by it. I think I'll definitely be taking a deeper dive into the understanding of this area and I encourage anybody else listening to this podcast to start thinking about the feasibility of taking the voice recognition as part of the healthcare put in your health care tool just because it's definitely something that is starting to become more commonplace. Jim, we're at the part of the show where we do the lightning round so it's a medical leadership course we're going to build the syllabus it's the 101 course or the ABC. On health care with Jim Harper and so I like to write out the syllabus with you four questions and then we'll follow up with a book that you recommend to the listeners. Sure. All right. What is the best way to improve healthcare outcomes?
Jim Harper: [00:22:27] I think one of the best ways to improve healthcare outcomes is to understand how health is changing with better information from the patient side not just from the by-products of care.
Saul Marquez: [00:22:38] What is the biggest mistake or pitfall to avoid.
Jim Harper: [00:22:40] I think the biggest pitfall to avoid and this is on my side as well as the healthcare side is believing that technology in itself is the solution to the problem. We need to be improving the relationships between individuals and the providers and the way to navigate the system.
Jim Harper: [00:22:57] And I think that voices is a great way to do it.
Saul Marquez: [00:23:02] How do you stay relevant as an organisation.
Saul Marquez: [00:23:04] Despite constant change.
Jim Harper: [00:23:06] I think that you stay relevant by trying to as Wayne Gretzky said skate to where the puck is going to be not where it has been so far but you try to anticipate the change in what consumers are learning to use and demand from other industries and how that may impact the healthcare industry.
Saul Marquez: [00:23:26] What is the one area of focus that should drive all else in the organization.
Jim Harper: [00:23:31] I think that the area of focus needs to drive us is improving the experience for the patients ultimately.
Saul Marquez: [00:23:39] And finally what book would you recommend to our listeners to read.
Jim Harper: [00:23:42] Yeah so in a funny way I am I'm a huge science fiction fan and although it doesn't seem like an obvious choice I think Ender's Game is a very interesting read and partly for me.
Saul Marquez: [00:23:56] I've seen the movie.
Jim Harper: [00:23:57] Yeah so I think the movie is good but one of the things that the movie doesn't emphasize is the role of ai Ender's Game has kind of a great character and relationship between ai and the main protagonist that's mediated by a device in New York called the jewel which is very much like a future version and that's why it's all of our intelligent voice assistance wannabe. And I think that in that dynamic we'll learn a lot about how seamlessly technology can integrate into the device and the more you reduce the friction to access information the more meaningful it is. So for me a lot of the takeaways and what I want my technology.
Jim Harper: [00:24:40] And what I think all to look like is that ever-present good in a true sense kind of advocate in your ear. So hopefully that's inspiring to some.
Saul Marquez: [00:24:49] Now I love that Ender's Game and so outcomes rocket listeners take note of these things. And the beauty of it is you don't have to take note.
Saul Marquez: [00:24:57] Go to outcomesrocket.com/jimharper.That's jimharper and you'll be able to see all of the show notes with the just links to the book that Jim recommended as well as links to his company and all the things that they've done. Check it out. Don't worry about taking note if you're driving running. That's the beauty of the podcast and the notes so be sure to check that out at outcomesrocket.com/jimharper. Jim, before we conclude I just want to ask you to share a closing thought and then the best way that the listeners can get a hold of you.
Jim Harper: [00:25:32] Yeah. So who a closing thought. I think that the hardest things to do are often the most worthwhile so as difficult as it is to change and improve outcomes I think it's something we all want to do.
Jim Harper: [00:25:43] And if you want to talk to me about ways to do that you can always email me at firstname.lastname@example.org
Saul Marquez: [00:25:51] Outstanding. Jim thank you so much. Looking forward to staying in touch with you.
Jim Harper: [00:25:54] I appreciate it.
Jim Harper: [00:25:55] Thank you very much.
: [00:26:00] Thanks for listening to the Outcomes Rocket podcast. Be sure to visit us on the web at www.outcomesrocket.com for the show notes, resources, inspiration, and so much more.