← Becker’s Healthcare Podcast
Becker’s Healthcare Podcast
Dave Newman, MD, Chief Medical Officer of Virtual Care at Sanford Health
00:14:53
Loading summary
Transcript17 lines
- [00:01]
A
All right, welcome, everyone, to the Becker's Healthcare Podcast. This is Giles Bruce. I'm an assistant editor with Becker's healthcare, covering health IT. Pleased to be joined by Dr. Dave Newman from Sanford Health. We just had him on a fireside chat at our AI Summit at Becker's CEO CFO Roundtable. So pleased to have him again for a podcast to talk more things virtual health and AI. And, Dave, can you please start by introducing yourself, telling me a little bit about your role and what you do at Stanford.
- [00:34]
B
Yeah, absolutely. So, I'm Dave Newman. It's good to see you again, Giles. I'm an endocrinologist and informaticist at Stanford Health, which is the nation's largest rural healthcare system. We're headquartered in the Dakotas. We serve a huge amount of territory in the Midwest. We've got hospitals, clinics, just about everything, every single subspecialty, and we are dedicated to being the leading rural healthcare system in the nation.
- [00:58]
A
Excellent. So, yeah, so I mentioned you did a fireside chat with us at the AI Summit, and we really were pleased that you were able to join us. But you asked a provocative question there. I found it interesting. Basically, when is it irresponsible not to use AI? So what indicators will tell us we've reached that point in healthcare? And are there any specific clinical areas? Insulin dosing, chronic kidney disease detection, perhaps, where you believe we're already close to that point.
- [01:27]
B
Okay, so let me start with a story about when I first started thinking about this. So, as an endocrinologist, I treat a lot of type 1 diabetes. So type 1 diabetes is a condition where your pancreas doesn't work, so you need to take insulin. And it, for years was an unbelievably hard condition to treat. Patients had to have a large amount of education to know their body, to know how many carbohydrates they're eating, to dose their insulin, when to take more, when to take less, when to eat more carbohydrates. And I had patients that I was seeing every three months that I could not get a handle on. Insulin pumps came out a long time ago, and they sort of helped, but still, the patient had to be armed with a huge amount of knowledge in order to continue to be successful with these. Even with the insulin pump and continuous glucose sensors which always check your blood sugar, patients were still poorly controlled because there was. There wasn't great communication. I have a patient. We'll call him Bruce. I had followed him for over 10 years, and I had not gotten him well controlled. That was on me. We had gone through everything. We had done diabetic education, he'd been a nutritionist. He knew everything about carbohydrates. But unfortunately it was a really hard condition and he was being admitted to the hospital for hypoglycemia or high blood sugars frequently. About two or three years ago, Bruce got something called a closed looped pump and sensor where it's an artificial intelligence enabled device that you wear in your belt that when your blood sugars go high it gives you more insulin and when it goes low it stops giving you insulin. This was mind blowing to both me and Bruce. All of a sudden it was like every eight seconds he had an endocrinologist sitting with him that could make decisions based upon how his blood sugars were acting. It would automatically give him more insulin. He only really had to be responsible for charging his pump and filling with insulin. Every three days. I have decreased my visits. I used to see him every three months, now I see him every six months. We might be able to go to every year visits because the AI is doing the work. It's better than I am. It took me realizing that I had to partner with the technology to do that with all my patients. As we move forward, we're going to see this with other conditions, whether it is in pathology, whether it is when we see peer reviewed studies where it's AI is better at diagnosing celiac disease. We have to educate our pathologists and they've got to be the ones that lead this charge. We want a human in the loop. So we want nephrologists, we want primary care doctors verifying that the CKD models are accurate. But at some point we have to partner with the technology. I kind of like the anecdote is when I go to my accountant, I would be really, really skeptical if they were using the abacus to do my taxes right. Like I want them using Excel. I want them using some sort of. I'm going to be skeptical if my doctors aren't using AI if it's the best available tool.
- [04:20]
A
Yeah, great example there about this technology really having some benefits for patients. You also reported near universal provider satisfaction with ambient documentation and even joked that you'd have a riot if it were removed. So why does ambient AI succeed where past documentation tools have failed? And, and how would you say it's changed the patient experience? You mentioned a moment you described when patients said it was the quote unquote first time you actually looked at them.
- [04:49]
B
Yeah, so I joked about having a ride now I think I would have a riot if we took away Ambient Experience. I get lots of emails, I get way more emails when our Ambient Experience tool goes down than any time else because it has improved so many doctors lives. I have got a buddy, we'll call him Sam. He's a pediatric oncologist, unbelievable doctor. But has never been good at writing notes. So he spent a lot of time with his, his kids and parents going over treatment plans, going over all the things that doctors do. But because of this, he wasn't great at entering everything into the medical record. He was always on what we call our naughty list. So overdue documentation. Um, after two weeks of having Ambient Experience, he was no longer on the naughty list. He came up and he gave me a big hug and he said I can't believe that I've got my life back. That he's able to see his patients do the things that he did. And it's easy. The problems with the old note taking tools, the problems with the other softwares is that it required lots of clicks, it required a somewhat unorthodox workflow that doctors had to do and it wasn't easy. Ambien is easy and it gives you your life back. Some statistics and in medicine we never reach 100%. 100% of the providers that have used this have said they want to keep doing it, they want to stay. 90% have reported greater job satisfaction and 95% have reported less mental fatigue. We firmly believe that this is going to be a tool that is going to help retain the best talent in the Midwest. That if we don't have these tools, you're not going to be able to recruit and retain doctors going forward.
- [06:31]
A
Yeah, yeah, great stuff. And yes, similar to some of the benefits I've been hearing from other leaders across the country on clinician satisfaction and burnout, et cetera. And next question. You gave an example at the conference of patients having to drive six hours for a 10 minute visit. Very powerful anecdote. So what would you say in the way of virtual care and telehealth is having the biggest impact on reducing some of that travel burden and miscare in rural communities.
- [07:00]
B
Yeah. So our statistic at Samford is virtual care saves patients an average of 176 miles of travel. That's just the mile saved. So it is time off of work, it is meals, it is hotels, it is putting your dog in doggy daycare, it is missing your kids gymnastics events. It's all those things that really add up that we're really bringing the Power back to the patients. One of my best examples is we have a phenomenal doctor in Fargo. His name is Seth and he's a bone marrow transplant doctor. He has revolutionized bone marrow transplant by leveraging virtual care. In the past, patients that did a bone marrow transplant required to stay in Fargo inpatient for five to six weeks because they needed so many cares. Now he's able to have them inpatient for two to three weeks and then send them home and do virtual visits for the rest of the time so they don't have to be in the hospital environment for that long. I've got lots of patients that I have to see every three months for Medicare requirements for diabetes. They no longer have to come to see me. They can do it from satellite clinics or they can do it from home. If now patients can get care close to home in the Midwest instead of going to a massive tertiary referral center, which they don't want to do, they want to stay with their doctors that know them and that have their records and know have been knowing them for over 10 years.
- [08:23]
A
Great stuff. Yeah. And I know from our previous reporting at Becker's, you know, how big of a investment and commitment that that Sanford has made to virtual care. It's turning to agentic AI, which is still, you know, kind of in its infancy in, in healthcare and elsewhere. But you mentioned some of your agentic outbound campaigns. Improving connect rates from 40% to 56%. You know, full 20 or so minute conversations with 16% of patients. Pretty remarkable stuff. What has surprised you most about how patients are responding to agentic AI and what do you think this has potential as far as, you know, closing care gaps for, for cancer screenings, primary care, chronic disease, etc.
- [09:05]
B
Yeah. I think the thing that surprised us the most was the satisfaction. So we are scoring over 8 out of 10 on patient satisfaction, which we never thought we would do. You know, from getting these, these bot calls, sometimes they can be quite challenging when it comes to health care. They know that we have are their best interests. Would I rather talk to a human sometimes? Absolutely. But would I rather talk to an agent agent to get my colonoscopy scheduled than not get my colonoscopy? Absolutely. That's the feedback that we've been getting. People want to close their care gaps. They want to get their lipid panels done, they want to have colon cancer screening, they want to, they want to have all these things done to make them healthy. The rate limiting step now is we don't have the workforce. So right now we're able to reach thousands more patients using an agentic agent and get them connected to the person that they need to get their, their visit scheduled or to get their specialty visit scheduled.
- [10:02]
A
Yeah, great stuff. And we'll definitely continue following that story is agentic AI evolves in the years to come. And you also spoke candidly at the conference about governance, separating clinical and operational committees, ensuring, you know, operators and clinicians also sit on each other's boards. What does good AI governance look like day to day at Sanford? And how do you build clinician trust without slowing innovation?
- [10:27]
B
So this is a story that happened to me out on call this week and I was getting lunch at one of our flagship hospitals in the doctor's lounge, and I literally had a doctor come up to me and say, hey, I've got this AI blood loss app. How do I use it? And I was like, that's super cool. Let's run it through governance. But he had never heard of governance. So it's up to us to communicate the strategy with everybody that has these products. We have to make sure they're safe and effective, which means that they have to be safe and effective through a doctor's lens. But there also has to be an operational lens, cybersecurity, that we've got to look at, that we've got to look at if it is fiscally responsible for the system to do. We've got, we've got to look at how we can have these apps work not just for one provider, but how do we scale it through the entire organization. We also have to look at how we monitor outcomes not only for the first six months, but going forward. How is this safe and effective for patients in the long term? We're talking about having real time dashboards that give insights into how these are trained and how they're performing. We need to make sure that the results are accurate for our patient populations, not just the patient populations of the the AI ventures were tested on. This is a lot of work and it requires a lot of people around the table, but we're very, very committed to doing what's best for the patient.
- [11:50]
A
Yeah, good stuff advice there. I know. Yeah. Health systems are definitely looking for guidance on their, their AI governance journey. So, yeah, some of that will, will definitely help our listeners. And your description of orthopedic residents practicing procedures with AI enabled virtual reality was also a highlight of our last chat. How do you see virtual training shaping the next decade of medical education? And do you think future clinicians will need a different skill set? Maybe being more like prompt engineers or workflow designers.
- [12:25]
B
I love this question and for the listeners, my mind was absolutely blown the first time that I saw the orthopedic surgery. AI VR, where it is, allows them to go through a surgery using virtual reality goggles and participate in the surgery before they go into the operator for a little bit of level setting. I have been a doctor for about 20 years and I learned from textbooks. YouTube wasn't really a thing. Even so, I couldn't even watch videos of surgery before I went in. We have evolved from textbooks, from videos and now to virtual reality and, and the learning is substantially improved. The orthopedic surgery residents can look at patient placement. They can see how the shoulder is supposed to be placed beforehand. They can map out anatomical landmarks. They can hit a button on their virtual reality goggles and the skin will be removed so they can look at musculature. They can hit another button, the muscles will be removed and they can look at vasculature. They can have the AI prompts have problem solving. So they can say, oh, you know, this vein or this nerve isn't in the right spot. How do I deal with that? I love your questions about being prompt engineers. So most like my kids now, I've got a college kid, I've got high school kids. They're amazing at using AI. They're great at designing their own prompts. Doctors need to get good at prompts to be better at doing medicine. This is something that we have to be educating on. One of the things that we see our virtual care initiative doing is being one of the nation's leaders on education, that we want people to be good at AI. And we've had lots of older doctors come to us and say, hey, I've never used these things. Can you give me a three to five minute video or sandbox environment to do this? So we designed our own curriculum to help people with that.
- [14:14]
A
Great stuff, as always. Yeah. With that we're going to wrap up, but really appreciate the talk today, Dave, and your insights on virtual care and AI. And also we look forward to seeing you. You're going to be speaking on a keynote panel at our Health IT Conference next next year in Chicago. So really appreciate you coming back to another Becker's event and doing the podcast today.
- [14:36]
B
I'm excited, Josh, thank you for having me.
- [14:38]
A
Awesome. We'll see you then, Dave.