Ep6: Dr. Uli K. Chettipally, MD, MPH, on the Need for AI in Healthcare

An interview with physician, author, and founder, Uli Chettipally

About Hills and Valleys

Hills and Valleys is a podcast that uncovers stories from leaders in healthcare, tech, and everything in between. Straight from the heart of Silicon Valley, we give you a look at the good, the bad, and the future; one episode at a time. Brought to you by Potrero Medical.

Listen to the podcast on any of the platforms below, watch the full video interview, or continue reading this blog to see the transcript.

About Dr. Uli Chettipally, Founder of InnovatorMD

Uli Chettipally, MD, MPH, is a speaker, physician, researcher and an innovator. He’s passionate about delivering artificial intelligence-enabled solutions to physicians at the point of care to improve patient outcomes.

Dr. Chettipally is the co-founder and CTO of CREST Network; a multi-center, collaborative research network in Kaiser Permanente, covering 21 hospitals in Northern California.

He designed, developed and implemented a region-wide clinical decision support platform to deliver real-time predictive analytics to the physicians – for which he received the “Pioneer” award from Kaiser Permanente Innovations. www.kpcrest.net

Courtesy of AI Med

Currently, Dr. Chettipally holds a faculty position as Assistant Clinical Professor of Medicine at University of California, San Francisco. He’s a Chairman of the Society of Physician Entrepreneurs, San Francisco Bay Area Chapter: a global biomedical innovation/entrepreneurship network.

Most recently, Dr. Chettipally authored a book titled Punish the Machine: The Promise of Artificial Intelligence in Health Care, which clearly explains the current health care problems facing the US and how AI technology can be used to decrease the burden on physicians, improve the quality of patient care, and decrease the cost for payers.


An interview with author and founder of InnovatorMD, Uli K. Chettipally

  • Interviewer: Omar M. Khateeb, Director of Growth at Potrero Medical
  • Interviewee: Dr. Uli K. Chettipally, Founder and President of InnovatorMD in San Francisco, California.

Khateeb: Hi everyone, Omar M. Khateeb, Director of Growth here at Potrero Medical and this is Hills and Valleys. We have another great episode for you today, with us is Dr. Uli K. Chettipally, MD., MPH. I want to give you a little background on him before we jump into it.

He’s a speaker, a physician, a researcher, and an innovator. His passion involves delivering artificial intelligence enabled solutions to physicians to help improve patient outcomes.

Now his work with the CREST Network in developing a clinical decision support platform earned him the Pioneer award for Innovation from Kaiser Permanente, which is right here in Northern California. He also received the Morris F. Collen award for his research with his team from the Permanente Medical Group.

To connect with him and learn more about his work you can also visit innovatorMD.com.

Khateeb: Also, he published a book called Punish the Machine: The Promise of Artificial Intelligence and Healthcare. Doctor, thank you for joining us.

Chettipally: Thank you Omar for having me here.

Khateeb: Absolutely. So, why don’t we start with your background? What influenced you to pursue medicine?

Chettipally: Well, growing up in India they were only two choices for us smart kids: either engineering or medicine. I’m not that great with numbers and I hated math, so I picked medicine. That’s how I got into medicine.

What I really wanted to become growing up as a child was a scientist. I wanted to become an inventor. It was very fascinating to me to look at machines and things that work that we take for granted, but don’t know much about. I was always very fascinated with machines and I wanted to become a scientist and an inventor.

Khateeb: Interesting. It seems that a lot of physicians see medicine as a way for them to extend the use of science and use science as a tool in a more practical environment. Did you see yourself in that regard when you were younger?

Chettipally: It was just fascination at first. Then I got into medical school, and I realized that so much of what we do in medicine affects people and affects patients. I realized how much good we can do through medicine.

It is one of those fields where you can make life and death decisions for the patient. You can change somebody’s life forever, whether it’s good or bad, and you can have a positive impact on society. And society itself looks up to healthcare providers and physicians to do this job.

Khateeb: When you were younger, who did you look up to?

Chettipally: My father was a great role model for me. He’s an engineer, but he had that inquisitive exploratory mindset. He would always question things that happen around him. I learned a lot from him about his philosophy of life and how to do good in society. More importantly, I learned how a single person’s contributions can have a major impact on society.

Khateeb: Wonderful. If you don’t mind me asking a rather personal question, what’s the most memorable thing your father ever told you?

Chettipally: There were many things, but one time I took some money that was supposed to pay my college fees, and I actually bought a drum set. A jazz drum set. It was expensive and a significant amount of money that my father gave be but I misused it.

Then later on he found out about it, that I didn’t pay the college fees, and so I went back and told him: “Hey, sorry Dad that I did this” and he did not blink an eye. He said okay, that’s fine, we will still pay the fees. And so he gave me the money. That totally, I mean, I knew my dad was cool and he was a very calm and quiet person. But that really blew me away.

Khateeb: Wow. How have you carried that lesson through life and, more importantly, into your work as a scientist, an entrepreneur, and a physician?

Chettipally: So I’ve incorporated some of those values into my own work life and family life. We’re all human beings, and when you’re human, you’re bound to make mistakes. That doesn’t mean that person is bad or that it’s always going to be that way. How you show grace and compassion to that other person – whether it’s a co-worker, your boss, or your patient – I think what makes a big impact on them is how you respond to those situations. That’s what I learned.

What’s the Current State of AI in Healthcare?

Photo by AI Trends

Khateeb: Well, you know on the topic of that, part of our show here is we like to start our listeners off at the top of the hill to get a really good glance at the state of things as they are before we get down into the valley. But, let’s start with AI since that’s what your book is about.

So where are we today with AI? Because some people think that AI is coming and other people have accepted the fact that AI is already here. What’s the current state of AI right now in the world and then more specifically in the world of medicine?

Chettipally: Lot’s of industries have gone through the digital revolution, where digitization of data has occurred much earlier than healthcare. Healthcare is one of the last industries that hasn’t been fully digitized yet.

Only recently, like about a decade ago, we started using electronic health records where we were able to capture data. I think for AI to be useful and provide value, it has a long way to go in health care. But in other industries, it’s automating work flows and giving insights into making decisions, whether it’s business or otherwise.

That hasn’t happened yet in healthcare, but we’re seeing signs of how useful AI can be because we’re getting data into a digital format. Now we can actually run tests and algorithms on it to see some of the insights that you can get.

What’s Hindering the Adoption of AI?

Courtesy of Modern Healthcare

Khateeb: So it’s really surprising to hear that doctor, because on the way to work I used Waze, which is basically a GPS that’s got an AI in it and gives you date and minute-by-minute updates. We use Google and YouTube search which has an AI in it to find the funny cat videos.

Why has an industry with the smartest and top performing people in a population taken this long to adopt AI?

Chettipally: So medicine and healthcare in general has this tradition of being analogs. A lot of healthcare that has been practiced is based on a personal connection, the trust between the patient and the doctor, and the compassion that a physician or a caregiver shows to a patient. There’s a big human element that is part of Healthcare. In fact, most of it has been that way for centuries. Only recently, maybe a hundred years ago or so, did we start looking at the science behind medicine.

So medicine was all art, and now we are slowly understanding the science part of it. Much of science today is driven by academic institutions or the industry, like pharma and medical device. This is mostly targeted towards coming up with new treatments and coming up with new drugs. And so the focus has been mostly on how do we come up with new care models or care ideas that can generate revenue?

Over the last 50 years or so, healthcare has become more of a fee-for-service type of situation where the cost of care is dependent on how much and how many clinic visits, how many operations, how many drugs you prescribed, and how many lab tests you order.

As we start collecting data, the real world data, through EHRs, we can see what we were doing in the past, and how that’s affecting the outcomes for the patient. We never had so much data available to us before to actually see what happens when we give drug A versus drug B versus drug C, or an operation versus no operation, or whether a test is useful or not.

So those kinds of evaluations we were not able to do in large populations using large data sets. Now, with the availability of this data through electronic health records, we have an opportunity to make a difference and to bring in new science, the data science piece, into healthcare. That is what is exciting about it.

When can we Expect Adoption?

Courtesy of Techburst.io

Khateeb: Do you feel that here in Silicon Valley, when we look at technology, we accept it as a part of life in terms of how it augments our skills and makes us better? Now that’s not always the case. When you leave the Valley you realize that we live within a bubble out here .

Out in the rest of America and the rest of the world, even though physicians and nurses and medical providers are under this kind of strain, do you feel like they are psychologically prepared to begin adopting technology and to understand that they are not capable of doing this all on their own?

Chettipally: I think that change will take some time, especially in healthcare. And yes, it is a complex field. But what we are seeing in other countries is that where there is a single payer and a single provider, which is the government in some situations like in England or Singapore and even in China, they are able to collect data on everybody and then they are able to change the way they were practicing medicine to be able to drive those positive outcomes.

We have a long way to go to become that, but at least we know that it is doable. It is doable and it is showing positive results, so the adoption will slowly kick in.

One big barrier is that healthcare in the U.S. is very scattered, there are lot’s of different entities dealing with data.

A disadvantage of that is that you cannot put everything together and then study that data. That’s why it’s taking longer for us to implement these technologies.

What’s Being Done with EMR Data?

Photo by Bricker.com

Khateeb: You mentioned earlier that healthcare is rather steadfast and conservative to pick up different technologies. Although it seems logical, it wasn’t that long ago that we finally started adopting and using electronic medical records. Now that we have an influx of data, what’s being done with that data right now at your average hospital?

Chettipally: If you think about electronic health records, it was not a voluntary adoption. They got pushed in and encouraged and incentives were provided by the government. So there was a lot of pushing and prodding to make that happen.

I’m glad it happened, but the problem is that EHRs aren’t designed to solve the problem of quality of care. EHRs are designed for, number one, bill for services. You document what you did and so that you can generate a bill.

They were designed to document what the physician or the care team did so that you can prevent malpractice. They’re designed to document patient care so you’re compliant with regulatory agencies when asked “Hey, are you doing quality checks?”

All that is great, but it does not solve the biggest problem, and the biggest problem is: how does providing care change for the better of the patient and the physician once you have done this? It does not.

Let’s say I saw a thousand patients with an ankle sprain, right? My next patient after 1000, my 1001st patient, will get the same treatment as the 1,000 patients before him. I can’t use the data collected on previous patients to improve care for the next patient I see.

So science kind of stagnated, although there was so much potential to improve care for the next patient that’s coming in. Now I took the ankle sprain example, but imagine it’s a very complex case like sepsis or heart attacks. Those diagnoses have a very complex way of dealing with the treatment, the assessment, the prognosis, and all of that.

Imagine if you collected all that data on those thousand patients. You now have the capability to figure out: what are the steps that were involved to result in a positive outcome? And can we use that knowledge on our next patient to make the care better and the outcomes better?

Shortcomings of the EMR System

Chettipally: A downside of the EMR or electronic medical records has been that physicians are forced to document things even though they know it’s not going to make any difference to the care that they’re providing. In a way, they were doing a lot of mechanical work and a lot of clerical work, getting frustrated as the fee-for-service system only compensates the work that you do.

When the compensation for each unit of work goes down, they started seeing more patients and doing more procedures. It is a very destructive force, and we see it as one reason why physicians are getting burnt out.

Instead of people doing that work and instead of physicians doing all their work, what if the machine does it? Will that relieve the work of the physician and the stress and amount of effort that goes into that? It definitely does.

And what if you automated a lot of those processes? Where a physician doesn’t have to click these boxes or check the boxes, but have the machine automatically record things and come up with preventive solutions and come up with suggestions for treatment options and suggestions for testing.

That would solve a lot of the problem, where you can actually get a better outcome using the machine learning capabilities and the artificial intelligence that this data can generate.

Khateeb: Your thoughts are that essentially, the EMR was introduced, then doctors and many of their colleagues became glorified number crunchers doing clerical work, or grunt work.

Nobody went to four years of medical school and 5-8 years of residency to spend their time doing that. You mentioned in your book that about fifty percent of physicians report burnout in the U.S.

Chettipally: Exactly. Physicians have a very high burnout rate and also high suicide rates.

Consequences of the Current Healthcare System

Courtesy of Health Impact News

Khateeb: Let’s talk about that, because it’s an ugly area of medicine that nobody wants to address. So a lot of Physicians report burnout. I think a report from the AMA four to five years ago stated that 49 percent of physicians said that they wouldn’t have gone to medical school and done it over if they had to do it again; that they wouldn’t go into medicine at all.

But let’s talk about the suicide rate. So there are doctors out there who are taking their lives because of the burnout. Can you tell us a little bit more about that?

Chettipally: In a typical fee-for-service system, let’s say you’re seeing 10 patients a day, and you get paid $100 a patient. And then somehow the payer or the insurance company decides that, “Oh we are paying too much. Let’s cut the rate to $80. That means you have to see more patients to be able to maintain your income.

And the regulations, the medical/legal risk, and also the documentation for billing purposes becomes more and more onerous with each year. And so they kind of start spinning their wheels really fast, and they have to do that to maintain their income. Because the rates have dropped per patient or per case or per test, physicians have to do more testing or issue more prescriptions or order more procedures just to maintain their income; and that becomes a very frustrating experience.

Physicians went into medicine number one to do good to the patients. Now some of the work that they’re doing they know to be unhelpful to the patients. And so that is a very frustrating endeavor.

Khateeb: Is this why you, in your book, mention the concept of “Spare the doctor save the patient”?

Chettipally: Yes. So right now we are making the physicians work harder and harder and harder. Instead of that, they should be doing more of what they do best: which is human interaction. Showing empathy, understanding the intricacies of disease, understanding the patient’s situation, building trust, being creative and coming up with creative solutions.

And that’s what human beings and physicians do best. But that’s not where they’re spending most of the time. Most of the time they’re in the number crunching and data entry stuff, which I think machines can do better.

Then on the other side, the patients are not getting the best of it either. Because if the physician is more involved in actual data entry and trying to get the billing right and the codes right, they’re not spending enough time with the patient to actually be helpful. And if the data is not helping the physician make the right decisions then that means the patient is the loser in this equation.

They’re not getting the right care, which they would have if the system was able to suggest the right treatments for this condition or the right testing based on how the outcomes will be changed, which is based on the treatment process. So patients could and should be healthier too.

Khateeb: The brighter future is to utilize the power of AI and the power of these computers and technology that we have to not only augment but actually help physicians and nurses avoid spending time doing the grunt work and instead spend more time healing.

The title of your book is Punish the Machine, can you tell us why we want to punish it?

Chettipally: Most of us know that the electronic health record, the EMRs and EHRs, have been developed with basic technology. Say, for example, technology from the 1980s and 1990s, that’s the technology that they use.

So in a way, we are pampering those systems. We are not getting the full value of the data that is in that system. Instead, we are punishing the doctors by making them do more work, such as entering the data. But the machines are very lazy right now. We’re pampering them and not expecting them to give us more information, do more work, or share more insights into the problem, so they’re not helping solve the actual issues with clinical care.

How can the outcomes be better for patients? Can we prevent disease? How can we postpone some of the illnesses that start very early and then become chronic diseases? I think machines have lots to contribute, but we have to take initiative to make machines work harder and smarter.

Accelerating AI Adoption

Courtesy of Dassault Systemes

Khateeb: It’s an interesting concept about how you get people to adopt technology. I don’t think that has changed in medicine. I believe that a lot of physicians and medical practitioners are still very reluctant, and even conservative about it.

We see this going back all the way to Ignaz Semmelweis, when he suggested you should wash your hands before dealing with patients, and it took 20 years for his colleagues to adopt that. So there’s this sort of bias, or rather cognitive bias, that exists. We cling to old paradigms that we are used to, even when new evidence comes along that challenges it.

So to avoid what happened to Semmelweis, in spending 20 years for physicians to adopt antisepsis practices, what can be done now by the community to help move this along faster?

Chettipally: I suggest changing the way research is done. Right now, it’s done through funding mechanisms and these funding mechanisms have their own priorities, whether it’s pharma or whether it is the NIH.

But I would bet you anything that if the healthcare company has any advantage, it is their data. If you can figure out how to squeeze insights from that data, it will benefit the company in the long run. That would be the best research because that is real world research.

Let’s say I have two million patients in my system and I’m monitoring their diabetes. I can figure out some things that research from outside, let’s say research done in Canada or research done on the East Coast, will not help with because my population is different. My population in California is very different from the population in the south, for example, because the risk factors are different, the level of diseases are different, and how they respond to treatment is also different.

So each healthcare entity has to come up with a plan on how they’re going to develop this knowledge base, with new knowledge coming out of their own data. They can do this through machine learning and AI, and then healthcare companies can use that information to make the outcomes better for their specific patients instead of relying on some published data from somewhere that may or may not be applicable to your real world situation.

Khateeb: Now as we know with human beings, data is really the last step in terms of influencing us, right? We become influenced primarily by the possibility of loss, the risk of missing out on something. How do you get your peers to…. maybe not adopt but to look at AI and higher technology in a different way?

Chettipally: There’s a big need for education. Most physicians don’t know much. I mean one of the reasons why I wrote this book is to educate people. There’s a problem where the way you study things and the way you understand what’s going on with patients is lacking.

Khateeb: Lacking where though? In medical school? Residency?

Chettipally: Yeah, it has just started in medical school and it should be a continuing process. One of the misconceptions physicians have is that, “Oh, if it’s research it has to be done in academic institutions. We don’t do research, we don’t care about research.”

Well, I think now it is possible to do research, because once you have the data you can do research and apply those findings into your practice very quickly.

Instead of waiting for somebody to do the work and then publish it and then it slowly goes into the guidelines, which may take 10 years or more for those findings to be implemented into practice, you can close that loop of research by testing, implementing, and looking at the outcomes. By doing this, the cycle continues and the value of that work is seen in the immediate benefits to patient’s lives.

Who Really Owns The Data?

Courtesy of HealthITSecurity

Khateeb: With regards to the research, I guess it goes back to data. Because in order to have machine learning, and later on AI, you need data.

So between the companies that create the technology to capture the data, and the hospitals who bring the patients in to create the data, and then the patient’s themselves who actually contribute to making the data, who owns the data at the end of the day?

Chettipally: Data has to be owned by the patient. When I say own it that doesn’t mean that they cannot designate other people to look at it or study it. Obviously hospitals are generating a lot of data from medical devices, but the ultimate pathway for that data to give insights is through the physician.

So the physician becomes the gateway to that knowledge, and is the ultimate decision-maker, all the while consulting with the patient and assessing what the patient’s choices are. And so it’s more like a shared decision making process between the physician and the patient, right?

Everything that happens outside of these two people, that knowledge and insight, needs to be funneled through the physician and patient during their interactions.

Envisioning the Future

Courtesy of Change Healthcare

Khateeb: So let’s get a little bit out of the valley and start going to a better place up the hill. Imagine that five years from now, physicians and medical practitioners have begun to widely accept AI and higher forms of technology that starts to free up their time.

What does that future look like? What do you see happening in the practice of medicine at that point?

Chettipally: I think physicians will be much happier, because lot’s of the scud work or grunt work will be done by machines. I think they’ll be spending more time interacting with patients. They will find out a lot more, and they’ll get a lot more insight from their interactions with the patient.

I think the patients will be happier, and I think the outcomes will be much better because they are based on data and based on evidence that is based on the knowledge that has been created through this process. I think there will definitely be positive outcomes that come out of this.

Khateeb: So you feel technology is going to free physicians to return back to the patient’s bedside. Do you think that there’s going to be an improvement in care because of that human interaction?

Chettipally: Yes.

When physicians are inundated with things they’re doing right now, they lose their capacity to think through the patient’s eyes.

Through the patient’s eyes meaning, what does the patient really want? Because it becomes very mechanical otherwise, and so now it is more of that human interaction, more of that trust and more of that engagement that helps them make the right decisions for the patient.

Khateeb: When you see patients come in, they are spending less and less time with their physician and nurse. Many times, especially when you’re sick, you feel very much like a victim of the disease. You almost can’t fend for yourself and rely a lot on the medical system.

Do you feel that having nurses and doctors spending more time with the patient will get them out of that victim mentality, and perhaps enable them to take a more active role as the protagonist in their healing story?

Chettipally: Yes, because the information that they get is more accurate and more reliable. As the physician spends more time with the patient, that engagement is where the physician is educating the patient, right?

They feel more empowered about their own disease processes, about the things that they can do to mitigate some of the bad risks and what are some things they could do to prevent undesirable side effects. So, yes, it’ll definitely engage the patients more and give them more knowledge and freedom that they can use to take care of things that don’t require a specialist or a physician.

Rapid Fire Questions + Closing Remarks

Courtesy of Machine Design

Khateeb: We’re very thankful for the time that you spent with us. But we have a few of what we like to call rapid fire questions. You can answer them as fast as you want or take as much time as you want with them. We will only do three today.

So first one: Your book is out on Amazon, and we’re going to have you back on the show to dive deeper into the book once our team reads it. But what book do you gift or give to people most often? Say somebody you mentor or a colleague.

Chettipally: In my career, one of the things I learned is that there are some things that I’m really good at. I didn’t know this at first, but what helped me was a book called Strengthsfinder. It is an evaluation that you can read, basically a book.

The premise is simple, that everybody is born with some strength. The sooner you figure out what your strengths are, the happier and more successful you’ll be, and the more you’ll enjoy your work life. So I know that one thing that I give to a lot of my students is that book.

Khateeb: Very nice. Do you have Netflix? What would be your favorite show on or off Netflix? Your all-time favorite.

Chettipally: I don’t know. I like the family comedies. I used to like Friends and it’s not there anymore. Simple stuff, that everyday stuff is funny and I like that.

I like one show, called In the Middle, where they have kids and I have kids so I understand the stuff that parents go through and simple things like that.

Khateeb: When you’re driving or commuting or traveling what do you like to listen to? Do you listen to music, podcasts, audible?

Chettipally: I like music. I listen to music mainly because it relaxes my mind and also lifts up my spirit.

Khateeb: Last question for you. Of all the platforms, what’s your favorite social media platform and why – if you have one?

Chettipally: Yeah, there’s only one that I actually use, which is LinkedIn. I see that there’s much better sharing of intellectual activity on LinkedIn, and so that’s the only one I use. I do have Twitter but I don’t use it that much.

Khateeb: Well doctor again, thank you very much for spending some time with us. I’m going to leave it in the show notes, but Punish the Machine: The Promise of Artificial Intelligence in Healthcare is on Amazon. Correct?

I’m holding the book right now, and I have to highly recommend it, especially to medical students and residents. Because you’re already spending a lot of time reading, but this book is literally only a hundred and thirteen pages so you can finish that in an afternoon.

It’s high yield and high value, it talks about the main things you need to know about AI and healthcare, and it’s a good time start learning about that.

So, we’ll have you back on the show in a few weeks, but for the listeners who are listening now, please go ahead and get the book and read through it.

When we have doctor Chettipally back on we’re going to dive much deeper into this topic to learn about punishing the machines and, more importantly, freeing physicians and nurses from the drudgery of data. So, thank you.

Chettipally: Thank you, Omar. Thank you for having me.