How AI and Automation are Transforming Healthcare

Brian Urban (00:22):
Yes, this is the Healthcare Rethink podcast. I am your host, Brian Urban, and today we are getting down with AI and automation in healthcare, and who else could join us to talk about this dynamic emerging space than the co-founder and current COO of CloudMedx, Sahar Arshad. Thank you so much for joining our little show today.

Sahar Arshad (00:46):
Thank you for having me here, Brian.

Brian Urban (00:48):
This is going to be so much fun. We're going to get to know CloudMedx more. We're going to get to know you more. Exactly how AI is maybe unifying, not disrupting the healthcare system, and really going to fix a lot of things going wrong. So, let's jump right into it. So, Sahar, thank you again for coming on the show. And with each episode we'd like to get our audience familiar with our guests, so could you just tell us a little bit about yourself and how you came to co-found this really creative AI tech startup that is having a really big impact?

Sahar Arshad (01:24):
So, my name is Sahar. I'm the co-founder and chief operating officer of CloudMedx, and I have a tech background. So, tech background, 20 years of experience in building products and multiple domains, hardware, software as well as healthcare. And around eight years ago we had a misdiagnosis in the family. And during that incident we figured that... A lot of times what happens, is that when a patient goes to a health system or a hospital, a lot of times there's either too much data and not enough time to go through the entire medical history or there's just insufficient data at that time where you have to make a decision, and patients fall through the cracks all the time.

(02:14):
And we started talking to a lot of doctors and we figured that it is a very common problem and something that can be partly prevented through technology. So, if you have a system in place that can predict adverse events in advance, highlight who the high risk patients are and provide recommendations backed by data to the provider at the point of care, it can prevent a lot of similar incidents. And that's how CloudMedx was born eight years ago.

Brian Urban (02:44):
That's amazing. And I'm thankful that you shared a little bit of your family experience, and I'm sorry to hear that it was challenging, but not an uncommon family example that I think a lot of great minded doers in our healthcare tech ecosystem have created different solutions and have wanted to take on a challenge because it's personally impacted themselves or their families. So, that's amazing. It's a purpose-based tech solution that obviously you're growing up. And eight years old now, so congratulations on going on now nine years... But eight years in the market.

(03:24):
And want to know more about your growing up story with CloudMedx. So, you entered into this space, not only being able to use data in an artificial intelligence framework, but you've added a lot of different solutions onto who you are as an organization. So, can you walk me through a little bit more about your development of CloudMedx and the services and solutions that you've really started to put out into the market today?

Sahar Arshad (03:52):
Yeah. So, the way the products and the company was developed because of big need. And so, even before we had a company, we had a customer that was asking us to aggregate data from different sources. They were an ACO that wanted to put all that information in one place and highlight who their high-risk patients were. And so, the first iteration of the product was population health

[inaudible 00:04:20] that could bring all the information in one place and highlight the risk stratification.

(04:26):
And then from then onwards to where we are now, we've worked with large health systems as well as payers to not only bring the information, the data, into one place, clean it up so that it's ready for downstream applications, to improve operations, to improve clinical outcomes, financial outcomes, as well as patient engagement. Because if you do clinical operational financial well, but the patient is not engaged, you would not get the right results.

(04:57):
So, it's a platform, but very modular, where you have number... There is basically five components. The first one is getting the data in, unifying it, and then four, applications on top of it. And it's backed by a lot of automation. So, everything that we do, we provide automation in it so that it's fast and it's efficient. And the problem that we see right now is that there's just a lot of manual overhead. Everything is done manually. You already have such a labor shortage, the two out of five healthcare workers are leaving.

(05:31):
And so, in this kind of market, if you're doing everything manually, it's just people are burning out. And so, I think that automation is the solution... Part solution to this problem where you can eliminate all the manual tasks and focus on, really, the patient connection that the providers can do. And there's just so much that happens behind the scenes like your documentation, putting in the information and extracting information, that can all be done through technology.

Brian Urban (06:09):
Sahar, how you hit on a couple really important things there that I think are often overlooked. One is patient centricity. You mentioned that up front. So, it has to be about pulling the patient into the journey. So, it's important to understand their experience and preference sensitive care needs that they might have, not just a whirlwind of automation around them, but how can it really touch them in a meaningful way. And your tech definitely does that. And connected to that, automation for the actual users. So, it's removing the administrative burden or manual burden that's on a lot of healthcare practitioners these days.

(06:48):
That has been, I think, the biggest challenge with the biggest opportunity for a lot of healthcare tech organizations over the last couple of decades here. So, it seems like, kind of, uniting all the healthcare stakeholders and aggregating the data, you were saying, in your modular platform can definitely be overwhelming. It's like, where do you start? There's so many uses. One thing I thought was really interesting about your technology is how it leverages Natural Language Processing, NLP. And I want to understand how that helps connect health plans and healthcare providers in an automated scenario. So, can you walk me through something that you use NLP for these days?

Sahar Arshad (07:33):
Yeah, so we use NLP in almost all components, which is the data aggregation part, as well as part of our applications. So, I'll give you two examples. So, when the data comes in, it comes in both structured and unstructured text. Unstructured text is like your discharge summaries, doctor notes, handwritten stuff. And a lot of that is not analyzed because it's not in a structured format. So, we use NLP to first normalize the data, identify if there's any missing information. For example, you want to do risk adjustment and you get a lot of data in and it's missing some important information such as patient enrollment information or some other data elements. It could highlight those missing data points upfront through NLP, so that right now what happens is, it's a very manual driven environment where you get a lot of data, someone manually scans through and it takes them a couple of weeks to identify this had missing information.

(08:33):
Then they go back, get the information, and then it's finally ready to do risk adjustment on. That's just one example. So, we use NLP over there and then in downstream applications as well. For example, you want to extract information from EMRs. Right now what happens is, nurses or highly qualified staff have to go in, extract information such as pain score and other information which is in doctor notes. So, it's a very manual process where a person can do five charts an hour. And also it takes them weeks to process or months to process hundreds of people, months to process a million records. Our technology, we did a recent deployment where the tech was able to do it in under five hours. So, that's the kind of efficiency... And we've seen in other industries where automation has really helped expedite the process and it's more efficient, it's more accurate, and then you can use the same people to do other work or process more data in the same amount of time.

Brian Urban (09:43):
And that's a wonderful example because in other industry, like you mentioned, manufacturing supply chain, it's worked really well in lean process management and then you apply that into healthcare and man, it's changing performance, quality of care and communication as well, inside a health service setting or even with a payer that might be affiliated or outside of a healthcare system. So, I love the example in particular with unstructured data that you had from physician notes. That has always been such a drag forever. And I think the more patients that are being seen on a smaller health center or hospital, it's just more work and more drain and more burnout. But when you offset that work, you can elevate the performance of the healthcare practitioner. So, it's a beautiful example.

(10:41):
And it kind of leads me to looking at some of the negative reaction that AI in particular... Let's step out of this example from NLP and go into AI a little bit here. So, it's been a wild evolution when we think about chat GPT-4 now, and even other industry leaders outside of healthcare have noted that there's a lot of danger with AI and it's come under concern from the US and from other organizations sitting outside of the US, but it can solve a lot of big gaps in communications. So, do you see more risk in AI in the future or do you see more deep benefits with AI as we refine how it's positioned and used?

Sahar Arshad (11:31):
Yeah, it's all about how it's positioned and used. You position it and use it the appropriate way and then you can use it for your benefit. I'll give you an example in healthcare. I think it's not whether we should use AI or not, it's a must have. There's so much data, there's like... More than 30% of the data is health data. So, it's not possible for a human to manually analyze all that data. So, you have to learn from these patterns that you can learn from so much data. Millions of data, then you can see what happened to similar patients, what kind of treatments work, what could work for this patient, what does a disease trajectory look like?

(12:10):
And you can really focus on preventative medicine. And so, without AI and technology, you're not able to do that. No one can. Our processing is limited, but a machines processing is much more, so you can do a lot if you use it, apply it appropriately. Of course you need to put in safeguards and make sure that it's used, and you have a physician in the loop, you have a person, the expert, in the loop so that they can validate everything and make the decisions. But it does make everyone's jobs easier.

Brian Urban (12:50):
I couldn't agree with you more. Obviously I'm biased toward that, but the application just makes sense. We have had such a big population change and growth in certain parts of the world that you can't possibly service all the needs that are happening or prevent things from happening that have adverse life effects if you don't use AI, if you don't use just a process of inserting it in the most appropriate ways and having safeguards. I think some of the comments, maybe fear-based comments, in the industry have been "AI is going to be used for autonomous surgical procedures or primary care," and things like... They're kind of leaping like 50 years into the future or 20 years in the future, rather than staying grounded with how we can learn how to use it together and not making up very odd scenarios. So, thank you for those examples because-

Sahar Arshad (13:44):
Yeah, we're still using paper forms, so there's a lot of opportunity to grow and put it in places where it can really elevate the experience for everyone. Patients, provider, payers. The experience is very, very broken. For a patient, there are long wait times, you don't know how much a procedure would cost, it's just a lot of confusion. And similarly with physicians it's the same problem. Lots of documentation and administrative work, so it can elevate the experience for everyone, even the administrators

[inaudible 00:14:24] where you have the right information at your fingertips and not having to wait for weeks to get the information, or not ever getting the data that you need.

Brian Urban (14:36):
Yeah, the timeliness of everything is really important too in delivering care and coordinating care. And you actually hit on a really interesting point that made me think about, there's a lot of processes that still use paper in a clinical workflow, whether it's rural, regional, large health systems, they still have some sort of paper, hard copy, physical manual process to them in a variety of settings. And I think we're starting to address all of these dated things we've done. And one big dated thing is trying to address social determinants of health and close inequities. And I love two things about your tech. One, you have a data visualization map on food access on your site, which is awesome and it's very helpful because that's a very big challenge in terms of addressing SDOH domains is food, is very, very challenging because it involves behavior and access and affordability.

(15:43):
So, we've really had health disparities like race, ethnicity, gender, language forever, since the dawn of time. And I think I feel like we're coming to a tipping point. And the access to social services is becoming more crucial now than ever. And I want to get your perspective on a societal cost. If the healthcare ecosystem, healthcare providers, systems, plans, pharmacy, pharmaceutical manufacturers, if they aren't investing into tech that is helping solve or advance health equity, what's the true cost? Like how bad is it going to get that we have to hit a wall before we invest into this space as an industry? I want to understand your perspective of, is it do or die right now? Or are you seeing some more adoption across the space with SDOH technology?

Sahar Arshad (16:42):
Yeah, I think there is a lot of interest these days. There's a lot of talk about health equity and I think they're headed in the right direction where there's a realization that these social determinants of health, like where you live, your income level, your education level, your access to food and nutrition, that plays a huge role in your clinical outcomes. So, what we do is, we highlight both clinical and non-clinical risks. So, these SDOH factors would be your nonclinical risk because, let's say a patient, Sahar, is not adhering... She's not taking her medications. And so, we know that medication adherence is a big problem and that can cost a lot to the patient, as well as the entire industry if the adherence is not in place.

(17:36):
But the reasons could be many and could be SDOH reasons. Maybe the person doesn't have a car or a transportation, or they don't have a pharmacy nearby or they're working two jobs and they don't have enough time. So, all these factors play a huge role in identifying those non-clinical risks so that they can be addressed and you can improve the outcomes and make sure that the patient is able to adhere to their plans and those factors are addressed. A lot of organizations have started giving Uber credits if the patient cannot... They have transportation issues, food delivery, medication delivery. And I think that is an area where people are putting in a lot of focus in and there needs to be even more focus there.

Brian Urban (18:27):
I love that you said that. So, that's a big call for our listeners. It needs to be even more focused on and invested into. And you brought up a really good point, you said "nonclinical." So, how important is it for health plans, payers in particular here, to have a view of nonclinical information? Basically everything that's happening outside of a doctor's office. How key is that for health plans to stand up programs or see what's going on in someone's life at the individual or group level?

Sahar Arshad (19:01):
Yeah, it's very, very important. So, they say more than 80% of your outcomes are determined by your social drivers of health. And so, being able to not only identify what those risks are and then do something about it, they should take an action and provide that accessibility to the patient. And also, go a step further and see, how do these social determinants of health impact chronic conditions, let's say. So, there is a lot of stuff that you can do by looking at what are the factors that could really, in a meaningful way, you can quantify how would that hit the bottom line and impact the outcomes.

Brian Urban (19:47):
And I think you know where I'm going too, because I wanted to get more information, not only on your perspective and your technology aimed at addressing social determinants of health barriers, but your actual tool, the SDOH Explorer. I find this pretty fascinating and I wanted our listeners, and personally, to learn a little bit more about the SDOH Explorer. And it's an incredible visual tool and I think we need that... Sometimes we... It's tough for us to just work off of reports and alerts, but it's better to use a tool that's visual and illustrative in nature. So, can you just give us an update on the SDOH Explorer?

Sahar Arshad (20:26):
Yeah, so the SDOH Explorer is... One version is available freely for everyone. So, you can go to SDOH Explorer, cloudmedxhealth.com, and what you can do is you can search any area, a county, even a zip code, and it will give you information about data, about food deserts,

[inaudible 00:20:47] security levels, income level, et cetera. And that gives you really... And it's color coded so you can see which areas have populations that may have a greater need for certain things.

(21:01):
And then we work with organizations where they provide their own data and then we can help them connect with organizations such as Meals on Wheels and that do the actual last mile delivery to really close that entire loop. And then we also provide a number of scenario modeling capability to see, "If you do this, how could that impact based on all the data that we've collected?" So, a public version is available, 3D, for anyone to go and zoom in and go into any zip code and explore. But then there's also another solution which goes behind the firewall where organizations can see their own member data.

Brian Urban (21:51):
And that's really cool, because I've used it and it's great to always test drive something, and you maintain it obviously because it can't be just static. So, I wanted to learn a little bit more about how you maintain the tool and then also, of course, any really exciting updates that you're having with any version twos or threes or fours with the explorer tool?

Sahar Arshad (22:15):
So, I added a lot of chronic condition data, over 20 chronic conditions data within the tool, so that you can see, how does it impact if you... Let's say, food insecurity, how does that impact diabetes? Let's say. So, a lot of fun interactive modeling that someone can play around and see how does it impact their population and what they can do about it. So, really the idea is not just to provide you a static dashboard, but something that you can take an action on.

Brian Urban (22:49):
That's really helpful, because I think that is giving a continual look for the user where I think a lot of older tools or publicly available data that is pulled for some platforms that are out there is really retroactive data. It's not really helpful, it's kind of in a timeframe in the past. But your tool is maintained and it gives an updated look for the users. So, for our listeners, check it out, SDOH Explorer, it's a really cool tool that's a publicly available version out there. And Sahar, I want to look at our crystal ball here and think about the next few years, maybe two years down the road, what do you feel CloudMedx's biggest contribution impact will be for the healthcare ecosystem?

Sahar Arshad (23:41):
So, in simple words, it's to elevate the experience for everyone that's involved. Patients, providers, payers. Because right now the experience is really broken. As I mentioned before, the long wait times, price transparency is a huge issue. Patients would get bills where they don't understand what it means, "in network, out of network," patient doesn't know all of that. Why is something costing them so much? And so, all of that should be transparent, easy for the patient, easy for the provider. It seems like non-trivial, it's not, but that's what I really want, is an experience... No one likes to go to the hospital, no one likes to get sick. And on top of that, the experience is so broken, so at least fix the experience so that they are on top of things. It's personalized to patients. Right now, if you buy something in retail, everything is personalized to us, to what we would like. I would like a personalized approach to medicine.

Brian Urban (24:46):
I love it. I love it. [inaudible 00:24:48].

Sahar Arshad (24:48):

[inaudible 00:24:47] with the technology that we have today, it's possible. It's just a matter of implementation.

Brian Urban (24:54):
And there's no excuse why we can't have this happen in the next five, 10 years, especially in our lifetime, for sure. But you're right, the healthcare experience is not great. It's fragmented and it's almost equivalent to going to a gas station experience. It's something you don't want to do, you just have to do, fill up your car, charge your car and get out. It's something that needs to evolve and change and it shouldn't be the last thing in our society to change because it's the most important thing that we have, is health and time. So, I love those final comments. So, Sahar Arshad, thank you so much for joining our little show today and for more excerpts and insights, please visit us at finthrive.com.

Sahar Arshad (25:44):
Thank you, Brian.

Brian Urban (25:45):
Thank you.