Podcast: Play in new window | Download
Today on HealthcarePittStop I have a very good friend with me, Khan Siddiqui. We’ve known each other close to a decade, we both have similar clinical training, and we’re both radiologists. Khan has had a varied career. Today we’ll talk about his work as an entrepreneur and his perspectives on healthcare generally. Khan, welcome.
Thank you.
I want to start with your background. You’re from Pakistan. What was training like for you to become a physician in Pakistan?
I’ll go back a little bit, about my role models and how I got to be a physician. Both my parents were physicians in the UK, so that started my thought process about what I wanted to do and what I was going to train in. Both being successful practitioners, they became my role models. My interests early on turned to technology. I started to do a lot of software development early on and did my first enterprise software solution out of high school. I sold that and decided to go to medical school. Medical school was interesting and very intense—an unbelievably awesome experience. Approximately half the faculty was trained abroad, so I had a very similar experience as I would have if I had trained here.
Sitting in Pakistan, you’d think the training is so different, but you realize it’s pretty similar to what you get here. The medical school I attended, Aga Khan University, gave me a lot of opportunities. I did my first EMR data-mining project over there. That was mining using a MUMPS database, which gave me an understanding of back-end technologies—that is what the Veterans Administration’s system is based out of; that’s what EPIC is based out of—so it gave me a huge edge when I came here because I was already playing around with similar databases on the back end. I also got the opportunity to work on code—on building predictive models for disease surveillance in the 90s—so this was way ahead of anybody else doing neuro-networks and AI type of work.
I remember you told me there were thousands of people competing for slots in medical school in Pakistan.
The year I applied there were about 4000 applicants and 865 made it. I know recent numbers are worse than that. It’s extremely hard to get into the med school I was in.
In these times with our new president, I worry that exceptional people like you at the top of the pyramid really won’t find their way to this country to add value.
You think of it in [terms of] today’s physician population: About 10 % of physicians are first generation immigrants. So the big impact this policy will have in healthcare is losing 10% of our physicians, or not getting those immigrant physicians on board. It’s really going to impact our healthcare delivery.
You went into radiology informatics—or computers and radiology—for several years, and you were really a leader in that field. You worked with Dr. Eliot Siegel at the University of Maryland, which has now become one of the leading departments in informatics. They had the first filmless department. You then left and worked for Microsoft and HealthVault. Can you tell me a little bit about what the goal was of Health Vault and why we don’t have Health Vault everywhere now?
At Microsoft I helped lead strategy for the medical imaging product line. When I got there, I became part of the engineering team, then took over execution on the engineering side of things, and eventually became part of the platform engineering team. So it was more on the product development and engineering aspect of the platform across those solutions. As part of that, I worked on different things also. But at the end of my career at Microsoft, they asked me to help out on HealthVault. We did a lot of interesting experiments to figure out what was happening, why consumers were not using it.
At the end of the day, we built amazing technology, and the philosophy was to build all the applications that partners brought in to bring rich experience on the platform. It really was designed as a repository for patients’ records, but our vision was that proper security, proper privacy, and delivering or engaging with the patient would come through our partners. But what the partners were thinking was that they were counting on Microsoft to bring the users in, because everyone was thinking there’s Hotmail, there’s X-Box Live, there were so many users, so let’s partner with Microsoft to try and bring those users in. No one really focused on bringing the users in. That’s what became challenging, to figure out how to acquire users and who was making it engaging for the users.
Healthcare is really hard. You don’t wake up every morning thinking about health—unless you are chronically ill and you have some active medical problem—so it’s really hard to do that [attract users]. Plus the noise that is out there for the typical consumer—the typical patient: they hear so much noise and marketing from the entities that are not healthcare related—from McDonald’s to entertainment—you are competing against that noise that is distracting the consumer. So it becomes really hard to engage in that aspect of it.
Health for the typical person creates a negative connotation. It reminds you of all the negative things—oh I had that burger or I had two scoops of ice cream instead of one—so in your mind you want to ignore it because it reminds you of all the bad things you’re doing. How to convert that into a positive connotation was something we struggled with. At the end of my tenure there, we figured out that’s what we needed to do, but the whole health solution group was spread out among joint co-chairs and the whole thing went a different route. HealthVault still exists; I know they’ve tried to do different things [but] I’ve not been in touch with the team to understand what they’re trying to do.
As an outsider, I look at HealthVault and almost every patient I know would like to have a digital shoebox, something to replace the paper records and the CDs they have. HealthVault seemed to be the ideal solution for that. But to me it never seemed to find a business model within the industry of health care to sponsor or promote it. Do you think that played a role?
I think it was more a consumer engagement issue. I’ll give you an example. Being part of the team, I made sure all of the medical records for our team and my family were on it. So one day my wife’s calling me from the kids’ school, and she needs all the medical records right away. I’m like, it’s on your phone, it’s in your hand right now. That’s when it hit me; I can’t get my own wife who hears about this all day long to think about opening the app on the phone and getting what she’s asking for. Instead she’s trying to get hold of me to go find the paperwork for her.
I can’t change that mindset, how, when thinking about healthcare, they need to open that thing where everything is. It’s a big challenge. How do we get the patient/consumer to think about that app when they need something? What we’ve learned from experience is that unless in your daily life you are doing something [with it], it doesn’t work. I spend a lot of time looking at all different apps and what the app “open decay” is, meaning how long it takes for you to forget about the app and never open it again. What’s that time duration? The T-half is 17 days. So if you have not opened an app within 17 days, the likelihood of your opening it on the 18th day is half. And that keeps going down. The peak is five days: if you keep opening the app every fifth day, you’ll keep opening it again and again. That’s what we’ve learned analyzing various app-open rates.
You once told me that you thought healthcare apps or devices needed to be a by-product of something pleasurable that you wanted to do.
That’s what I firmly believe. All the variable stuff needs to be a fashion accessory first, or some other accessory that you wear because you’re enjoying it, not because it’s a health tracker.
[That function] needs to be secondary. Do you think things are going to change now that hospital systems and payers are going at risk, and they’re really incentivized to make elegant healthcare solutions?
Without hitting the right incentive for the patient it’ll be hard. At the end of the day, that’s the person that needs to care. If the patient’s not engaged, no matter how good your product is, it’s not going to work.
You told me how you were involved with X-Box and how some of your radiology helped their product.
We started working on how to recognize human anatomy in medical images using deep learning in 2008, before IBM Watson or anything existed. We released our medical imagining deep learning product in Microsoft in 2011 and got through FDA approval also. As part of that work we built technology to recognize human anatomy—algorithms recognizing anatomy at scale—and that’s what ended up becoming the foundation technology for X-Box Connect. Our focus was on building the algorithms recognizing the anatomy and any kind of imagery. There are a lot of different aspects to it, and the first [consideration] when we were starting this whole thing was to actually do deep learning—in those days deep learning [was] on GPU’s (the type of processor used for gaming applications). Two people on my team were trying to do 3D visualization, and there was a team that thought CPU (more standard computer chips) was the right way and a team that thought GPU was the right approach. And the CPU guys always won because any time they saw any slowness they’d just slump up another CPU and it was faster. It was harder for the GPU guys to create forms of GPUs to do it. So in one of our efforts for folks who were going to experiment, one of my GPU guys decided to rewrite the entire medical imaging codec in machine language to remove all the accessories to make it much faster. So they were able to mak
e it 200 times faster than the CPU, but then it became a problem for CPU to scale that much, so what they did was get an understanding of how to code for a GPU. That’s how we started writing the codecs for people learning in GPU.
There’s this mash-up now between the gaming technologies and a lot of medical technologies. Some of the stuff you did as a radiologist makes it possible for us to sword-fight on the X-Box to be able to separate two combatants, which is amazing to me.
Those apps are amazing. [Next time, Khan] I want to shift a little bit and talk about your new company, Higi.
Reader Interactions