Latest / CareTalk: Healthcare. Unfiltered. / Should Nurses Feel Threatened by AI?

Transcript

Should Nurses Feel Threatened by AI? · CareTalk: Healthcare. Unfiltered.

• 0:00
  Nurses in San Francisco are protesting the use of artificial intelligence in healthcare.
• 0:05
  Nurses elsewhere are complaining too, claiming that untested, unregulated technology is making their lives harder and endangering patients.
• 0:13
  Do they have a point?
• 0:14
  Or are they just holding back progress to protect their own financial interests?
• 0:22
  you Welcome to Care Talk, America's home for incisive debate about healthcare business and policy.
• 0:31
  I'm David Williams, president of Health Business Group.
• 0:33
  And I'm John Driscoll, senior advisor at Walgreens.
• 0:36
  Everyone deserves healthcare that helps them feel and live their best.
• 0:40
  That's why TriHealth is breaking down the barriers that stand between patients and living a healthy future.
• 0:46
  With five hospitals and over 140 care sites, TriHealth is making it easier for Cincinnati area residents to stay proactive about their health.
• 0:55
  get the care they need and be seen, heard and healed.
• 0:58
  To learn more about how TriHealth is transforming care through their groundbreaking population health approach, visit www .trihealth .com.
• 1:07
  David, I think that nurses are genuinely concerned about whether the race to win the AI battle as hospitals embrace new technologies isn't...
• 1:24
  is moving so fast that it is leaving the critical caregiving role of nurses behind.
• 1:32
  I think it's the nurses are protesting and are concerned about genuine issues here.
• 1:38
  You sound really skeptical.
• 1:41
  Well, John, it's just my intro.
• 1:42
  You know, sometimes people ask me why I'm so defensive and I say, well, let's get people are always attacking me.
• 1:46
  So that's part of it.
• 1:49
  Well, nurses, my view is that they're overly concerned about.
• 1:54
  AI and that AI can actually do a lot of wonderful things to free up time.
• 2:00
  You know, when people are in a hospital and they press the nurse call button, first of all, sometimes they're afraid to push it because they're worried the nurse is going to be upset with them.
• 2:08
  And if they do push it, then, you know, the nurses may be busy and can't even get there.
• 2:14
  And when and when they do, you know, they don't have time for the patient or they didn't understand them completely.
• 2:19
  So I think that, you know, the.
• 2:23
  the nurses are concerned because maybe they know they're not providing the type of service that the patient actually needs.
• 2:28
  David, David, David.
• 2:29
  So let's define our terms here.
• 2:31
  First of all, nurses are the critical, caregiving, evaluating bedside partners for patients in their care journey.
• 2:39
  Everyone, most people at least, understand that nurses are kind of critical to healing.
• 2:46
  They're the sentient and heart -centered layer.
• 2:51
  of care in America.
• 2:52
  Doctors are just always, there's not enough doctors and not enough time to do all the work that doctors do.
• 2:58
  So nurses are at the bedside.
• 2:59
  So they're pretty critical in the healing process.
• 3:02
  And I think that what they're, there's always people like you who want to embrace technology very, very quickly.
• 3:11
  And I think that's admirable in a lot of ways.
• 3:13
  But I think we should heed the call for really concern, not stopping.
• 3:20
  that nurses are organizing around, particularly at Kaiser Permanente and in the West Coast where there's been a lot of noise about this, to make sure that the intelligence and the workflow advances don't actually eliminate or marginalize their critical caregiving role.
• 3:42
  I mean, what say you to that?
• 3:45
  Well, Johnny, you know, you enjoy painting me as an enemy of nurses, and I hope that I don't come across like that.
• 3:50
  Let me explain why there's a couple of good reasons and why I think this is coming up sort of early on in the technological transportation transformation.
• 3:58
  I think a lot of it points back to what happened with electronic medical record.
• 4:01
  So electronic medical record, you know, great, we're going to digitize the workflow and so on.
• 4:06
  But what ended up happening was that the people who are administrators and scientists and computer programmers.
• 4:13
  You know set up electronic medical records for what they wanted to see what they thought would be good without an appreciation of what actually occurs you know at the bedside with nurses who are the ones that actually enter information to electronic medical record now nurses they have to get stuff done right so what i was saying about call buttons and all that they really have too much to do and so they have workflows and guess what they have workarounds so that things actually happen and if you go and replace that.
• 4:40
  or you try to design some artificial intelligence in without understanding what's going on, there's going to be a problem.
• 4:46
  So I think number one, nurses want to be involved.
• 4:48
  That makes good sense.
• 4:50
  And I think that there are places where then the AI is coming in and it's not just like artificial intelligence, but it's like some nosy obnoxious person coming in and saying, hey, do this, do that.
• 5:01
  And it's useless.
• 5:02
  So for example, you can use AI and you can see a great conference presentation, John, at some of the famous conferences you like to go to and it'll show, look.
• 5:09
  it predicted when someone's heartbeat was going to change and it was going to be a problem.
• 5:12
  Well, what happens is that those anomalies then come and they come to the nurses like just all these extra alerts that are false positives and they go investigate them, which first of all, took time away from the patients.
• 5:23
  Genuine alert fatigue.
• 5:25
  Yeah.
• 5:26
  And then what happens when there are so many alerts based on predictive algorithms or false signals.
• 5:33
  is that doctors and nurses start not paying attention to them.
• 5:36
  I think that's a legitimate issue, but maybe we break it down into different pieces.
• 5:40
  I think your comment about electronic medical records are the databases that every doctor, hospital, and patient's office use to log and then bill for and then identify what clinically is going on with a patient so that information can be tracked.
• 6:02
  The theory behind that was great.
• 6:05
  And to your point, getting the electronic part of the records away from paper allowed us to digitally really track patients over time and make sure that they were clinically accurate and then build correctly and hopefully that patients wouldn't fall out of the system.
• 6:25
  That was a great idea.
• 6:27
  But there was a major design flaw in that all of this data entry and user interface work that was supposed to help clinicians track patients better often ended up just absorbing more and more of their time.
• 6:45
  So they're spending less time with patients and more time with effectively typing stuff into records.
• 6:51
  And what that that's created is less actually patient and nurse and doctor interaction.
• 6:58
  I think that's that is one of the concerns here that you know, hundreds of billions of dollars went into this, you know, like, you know, sort of turning on the electronic medical records and creating a digital age without really necessarily improving care.
• 7:14
  I mean, it's unclear yet that we really realized all of that.
• 7:18
  And I think the nurses are really pushing back on that.
• 7:21
  And I think I think that's reasonable because, you know, there's a there's a lot of lessons in user centered design of any sort of product, but particularly digital products that you want.
• 7:32
  that the kind of the critical players to be involved in the design and administration of new technologies or they will be designed in a flawed fashion.
• 7:41
  I think that's a lot of what's why nurses are pushing back.
• 7:45
  Well, John, I think when you mentioned Kaiser and San Francisco, I think a couple of things that are happening.
• 7:51
  One is that they're in San Francisco.
• 7:53
  So there's a lot of these tech bros out there that are going and driving technology pretty fast.
• 7:58
  And it's pretty clear that some of the things that they're putting in place are immature, not just those individuals, but the technologies too.
• 8:05
  And so you get pushback against the first generation.
• 8:07
  And my concern is that they'll be, you know, we'll have too much of a delay in bringing AI in, which is actually pretty, can be pretty useful for nurses just because of their in that, in that specific situation.
• 8:18
  Let's talk about some of the ways that AI can actually be helpful for nurses.
• 8:21
  Before you go there, maybe David, we could just explain what AI is because there's sort of a.
• 8:27
  a man behind the mask or a man behind the screen aspect to this, artificial intelligence is kind of a scary phrase in many ways.
• 8:35
  Maybe we, do you want to just explain what artificial intelligence is and how it's, how people are starting to think about bringing it into workflow and digital workflow?
• 8:46
  So artificial intelligence is something that simulates human intelligence.
• 8:50
  And so it can appear to be very much like a thinking individual.
• 8:55
  And so that's what makes it I'm kind of scary.
• 8:59
  It's also it makes it very powerful.
• 9:01
  The definition of artificial intelligence actually and what is considered that actually changes over time as we start to understand what the technology is.
• 9:08
  And some say, well, you know, the current generative AI was just linear algebra, you know, so it's it's things, it's technologies that are simulating what a human is doing and doing that better and better.
• 9:20
  And it may get to the point of having superhumans that it's doing things a lot better than anyone.
• 9:25
  as a podcast partner here?
• 9:28
  You want to replace me?
• 9:30
  I'm just asking.
• 9:32
  Because I think really the way to think about AI, the way I think about it, and I think you did a really nice job describing it, is with all of the advances in the ability of computational power, the ability to calculate massive amounts of digital information, whether it's visual through computer vision or through numbers, the zeros and ones are through words.
• 9:57
  It's allowed us to absorb all of that information and then query and improve a lot of human workflows and inquiries at a massive scale so fast and so compellingly that at times it does, the goal is to mirror the ways people think, but to be able to have a machine that could do it much faster, much more accurately.
• 10:24
  to extend what we think of as knowledge and understanding today.
• 10:29
  And so where I think that applies in healthcare is obviously around things like digital exploration of new pharmaceutical compounds.
• 10:40
  But it also is directly relevant to things like just paying bills more effectively, creating an explanation of benefits that someone from an insurance company that actually any patient could understand.
• 10:51
  There are a lot of applications, once you can find the right AI tools to apply with the right information that could potentially improve a lot of things in your, you know, the conventional form of AI is this, you know, chat, GBT, OpenAI, and all of the large technologies are, you know, talking about a consumer interface.
• 11:12
  But what's interesting is that AI, which is really just think about it as big compute, is going to be applied, I think, to every aspect of healthcare workflow, David.
• 11:21
  And I just think that the, you know, that engaging the users in that is pretty critical.
• 11:27
  But let's get back to your having given people a sense of what it is.
• 11:31
  What's the next vector of your defensive explanation of this issue?
• 11:37
  Yeah, what's the vector, Victor?
• 11:38
  So let me concede that nurses are it's reasonable to be pushing back and asking questions.
• 11:43
  But let's talk about some of the things that AI can do for nurses.
• 11:47
  So one of them is actually to get the value out of the electronic medical records.
• 11:51
  That is you were saying before, maybe hasn't been realized to date.
• 11:53
  So one of the problems with electronic medical record is now instead of the nurse looking you in the eye and hearing you and really seeing you and understanding you, they have to type a bunch of stuff into the keep and with a keyboard.
• 12:05
  So one of the things AI can do is ambient listening can be listening in and also computer vision.
• 12:10
  So you could basically get computers out of the rooms.
• 12:12
  and actually have it so that you've got an assistant that's there listening in a way and the nurse doesn't have to spend all the time and documentation.
• 12:19
  Another piece is just sorting through the medical records.
• 12:22
  So there's all this information that's captured, but someone comes into the room, they only have a few minutes to deal with it, and it's hard to get the essence of it, which is what AI can do well.
• 12:31
  Another issue is that when nurses are changing shifts, for example, there's this transfer of information from one shift to the next, and it's usually pretty superficial.
• 12:41
  just given the amount of time that's available.
• 12:44
  And sometimes things are missed.
• 12:45
  And when there's medical errors, it often has to do with this.
• 12:47
  Well, AI can be pretty helpful in terms of improving communication.
• 12:51
  And then the last one I'll mention for now is about education.
• 12:54
  So there's not enough nurses or it's hard for nurses to do the continuing medical education because there's a teacher shortage of nurses.
• 13:01
  And AI can actually be quite helpful on the educational side.
• 13:04
  You're just marrying me with information.
• 13:05
  But I understand all of that.
• 13:08
  But aren't we really talking about a common and not unreasonable nervousness that people aren't being engaged in the design of this.
• 13:21
  I mean, you can come up with all these examples.
• 13:23
  I mean, everybody came up with all these great examples of electronic belt crackers.
• 13:26
  I'm not here to contest that more computation isn't better.
• 13:31
  But I think what nurses are genuinely worried about, what doctors and patients are worried about, is the first use case going to be to reduce labor, simplify bills, and maybe put the machine between themselves and the caregiver.
• 13:48
  I don't think that's going to happen.
• 13:50
  But I do think that's a legitimate concern.
• 13:53
  There's no question that at a time of an aging population and limits in the doctor and nurse workforce, older people need more care.
• 14:03
  They are going to need more access to nurses and doctors.
• 14:06
  That if the The machine can be an extension of them.
• 14:09
  It's a positive.
• 14:11
  If it's a replacement, I think I'm kind of nervous.
• 14:15
  Yeah.
• 14:15
  All right.
• 14:15
  Well, you're kind of a nervous Nellie anyway, John, but there's drugs for that.
• 14:19
  Luckily.
• 14:20
  Listen, we also have a big problem, which is the cost of health care.
• 14:24
  It's huge.
• 14:25
  It's rising fast.
• 14:26
  And hospitals are very expensive.
• 14:28
  And the biggest expense of a hospital is the nurses.
• 14:31
  And we don't have a lot of immigration.
• 14:34
  There's more demand for nurses.
• 14:35
  And guess what?
• 14:36
  If you want to contain the costs of healthcare, especially the cost of hospitals, you have to do something to address the cost of nursing.
• 14:44
  Now, how are you going to do that?
• 14:46
  Well, one way is to provide more capital in a sense, you know, per nurse so that they can become more efficient.
• 14:52
  And so there is a genuine sort of, you know, concern about the patient and so on.
• 14:59
  And a fear about cost reduction.
• 15:01
  Well, There's got, you know, something's got to give somewhere.
• 15:03
  So there's, there can be multiple things happening, right?
• 15:06
  Which is the nurses say, Hey, there's terrible things happening.
• 15:08
  We should be involved.
• 15:09
  But also, yeah, there is some, let's hope somebody's paying attention to costs.
• 15:14
  And if you're going to look at costs in a hospital, it's got to include nurses, John.
• 15:17
  So there is attention.
• 15:20
  Well, I think that's fair, but I mean, all, all, all the nurses are asking for all the doctors are out.
• 15:27
  The patients want is to be involved.
• 15:29
  in the design of these things.
• 15:30
  There's no question there's a tremendous amount of administrative trivia that you'd love to automate, simplify, and give to the machine.
• 15:36
  And there's a lot of things around billing and logging and tracking that the machine can do.
• 15:41
  I just think people just want to make sure that the machine doesn't get in the way.
• 15:46
  I do think you raise a really important point, which is that we are at some point not going to be able to afford the total cost of care, and we're still going to have labor shortages.
• 15:56
  And so I think that is a reasonable thing, but I think it's more about taking the simple and the stupid out of workflow and not replacing doctors and nurses, which is, I think, what you're, are you suggesting that?
• 16:11
  No, John, I mean, I'm suggesting reducing nursing costs in total, keeping the growth constrained.
• 16:20
  We need to look at costs.
• 16:21
  Let me turn to something that is an example.
• 16:24
  What I want to see is, I think that AI is a potential way, to try to break this compromise, you know, the iron triangle between, you know, cost, quality and access.
• 16:35
  So let me give an example that's outside of the realm of what's being talked about now, but where - Can you just kind of step back and say, what is your iron triangle?
• 16:44
  It sounds very intimidating.
• 16:45
  Yeah, it does sound intimidating.
• 16:47
  It's like the Bermuda triangle, except that one's not real and the iron triangle is real.
• 16:51
  So if you want in healthcare, you want to have high quality healthcare.
• 16:55
  at a reasonable cost and you want people to be able to access it.
• 16:58
  Well, the next thing is usually pick any two, right?
• 17:01
  Because if I want to have low cost and high quality, it's probably going to be by limiting access and having a lot of waiting lists and so on.
• 17:10
  So that's the iron triangle says you can't really get away.
• 17:13
  You can only be on one of the sides.
• 17:16
  Yeah.
• 17:16
  OK, so maybe this is a way to to get rid of that, right?
• 17:19
  Because you can reduce costs at the same time.
• 17:21
  You're keeping quality up and you can improve access.
• 17:24
  Let me give an example here of something we've talked about before.
• 17:27
  Violence against hospital staff, John, nurses, doctors and so on.
• 17:32
  Right.
• 17:32
  We talked about this big problem.
• 17:33
  Well, one of the costs of health care now, of course, is that you've got to have more security.
• 17:37
  Pay people for that.
• 17:38
  And also you're going to reduce productivity if you have to be worried about violence occurring against your staff.
• 17:44
  Well, AI can help potentially.
• 17:46
  So what can it do?
• 17:48
  Predictive analysis to say, where is there going to be an issue?
• 17:50
  Surveillance and monitoring, which also, you know, multiple sides to that.
• 17:54
  simulations and training, personalized interaction with patients.
• 17:58
  But AI can be applied to a problem like that in a way that's better than just more security officers, metal detectors, fences around the building, ID checks, and so on.
• 18:08
  So we can use AI to get at some of the things that really matter, and we can bring the nurses into it and other staff and address their concerns.
• 18:17
  I think that's a phenomenal use case.
• 18:19
  Finally, you're suggesting something that I think we can both agree on David, because the post COVID, we really do have a much bigger problem of violence than we've ever had in hospitals.
• 18:31
  And really the respect for doctors and nurses, sort of the invisible bridge between that and the fear that most people, that many people feel when they're in difficult circumstances seem to fall away.
• 18:46
  And there's much more violence against.
• 18:48
  doctors and nurses in hospitals every day.
• 18:50
  If we could actually use algorithms to keep them safer and predict who's most at risk of creating that kind of a violent incident, I think that's something that we would all agree on.
• 19:01
  And honestly, it's not a use case that I've seen many AI companies approach.
• 19:07
  So maybe we can provoke some interest in that, because I think every hospital in America would appreciate getting that right.
• 19:16
  Well, John, I love to ending on a point of unity.
• 19:18
  So let's do that and say that's it for yet another episode of Care Talk.
• 19:22
  We've been talking about artificial intelligence and nurses on the front lines.
• 19:27
  I'm David Blams, president of Health Business Group.
• 19:29
  And I'm John Driscoll, senior advisor at Walgreens, friend of nurses.
• 19:33
  And if you like what you heard, you didn't, we'd love it if you subscribe on your favorite service.