Chyrill Sandrini: Welcome back to HTM Insider. Thanks for tuning in today and listening. My name is Chyrill. Sponsored by MultiMedical Systems, we bring this podcast to you every month. We're happy to report that we are just getting more exciting guests every time we turn around. I'm so thrilled today to have on Pete. Pete's with Moffitt Cancer Center and the Director of Enterprise Technology. And we're going to
talk about AI today. Now, some of it scares me. Some of it confuses me. And we hope to like dive into how it fixes, maybe your addresses problems in the healthcare technology fields. So Pete, why don't you introduce yourself to our listeners? Pete Daddio: Sure, thank you. So I'm Pete Daddio, I'm the Director of Enterprise Technology, as you stated, at Moffitt Cancer Center, down in Tampa, Florida, nice and balmy, just getting through hurricane season here, which certainly offers challenges. And I've been in it for probably 20 plus years
or so. And certainly seen a an interesting lexicon of how technology has changed over the years, and especially when I started right in the late 90s. So and now to AI. And to your point before AI is scary. It scares me too. Chyrill Sandrini: Yeah, you know, gosh, I'm gonna date myself a little bit, I think you're going to kind of be with me on this. I remember thinking artificial intelligence was C3PO, R2D2. That's what I mean, originally, when it first came
out. We're like robots can talk and move and even though it's on the big screen, but now we can do that. Right. But according to what AI is, to the healthcare technology field to healthcare in general. Let's dive in on that. Like, what are we using it
for currently? And maybe what you're using it for personally in your facility? Pete Daddio: Sure. So I think AI is really starting to hit from a healthcare perspective, in the sense of what are all the different things, how AI can help, how can that help augment? And I think there's a lot of trepidation. Because there's a lot of ethical questions. However, what we're trying to get to is, let's not replace everything with AI, let's go down and find things and find processes and find ways that AI can be very helpful. And we're starting to see some of this in our clinical space. And there's a lot of stories of just clinicians and nurses and other clinical professionals that are just overworked and spend so much time with patients and spend so much time with patient records. So some of the areas
that our facility and many other healthcare organizations are how can AI help Clinical Professionals be better or more efficient with things like clinical notes and patient records? Dictation dictation has been a dictation of notes has been a thing for quite a while. But how do we get it into those medical records? How can it help augment to say if you know, there's Pajama Time, which is where clinicians after dinner usually at home will spend hours upon hours doing dictation of notes and updating cases? How can we bring that down to a shorter time? How can AI help with that, and maybe Hey, during a clinical or patient session, those notes that are taken, maybe they can be added into a record using AI and then a doctor or nurse can go in after and just edit make sure you know that's exactly what we needed, or hear some of the edits, and really take that time down from hours to maybe say 30 minutes. And some of the other things too are how do we help with the patient experience? What type of technology can we use to help augment some of those patient activities and make the patient experience much more valuable, and especially in the medical field and here at Moffitt Cancer Center? When we have a patient, it's a really traumatic and potentially grieving experience when it comes to cancer. So how do we take that extra step to
make that patient experience as comfortable as possible, not just for the patient, but also for the caregivers. So we're starting to use technology, and to help the patient experience there as well with some of our smart rooms. So those are a few of the activities that we're looking at right now and starting to implement.
Chyrill Sandrini: So if I'm a patient and I hear of AI, and maybe I'm not tech savvy, and thinking that the doctors taking notes, with AI behind it. I know that right now, every time I'm typing something and I'm in certain platforms, it said would you like to respond us In AI, and I will sometimes I'll click it, I'm like, Ah, that's not at all what I said would say. So as a patient, how should the patient feel about it? Should they feel comfortable? Should they feel that this is accurate? And is it secure? Pete Daddio: Very good question. And I think that's where the ethical part comes in, in the sense of putting the brakes on certain activities, but also, let's assess where that right sweet spot is, if you will. And I think what we're seeing is with clinical notes that they're not exactly input into the patient router, record, automatically, what happens is, let's create a summary of these notes, and then have that clinical professional older, its nurse or a doctor, review those notes and then approve those notes. And I think that is one
way that we're starting to see some opportunity of a really expanding in that space, but in the sense of, the doctor still has to approve it, or the nurse still has to approve it. So it's not to your point, with having translation happening, there will be bugs, there will be words that are not translated properly, but then that gives the opportunity for that clinical professional to review those notes, to be able to either save, edit, change, and then be able to enter them in. But it's still the time saver. So I think we're looking in those spaces. And then also working with patients, if there are those types of questions, we could still do it, quote, unquote, the old fashioned way.
Chyrill Sandrini: So when you say these notes, does an AI need the information to be able to generate it? And how is it being fed into that stream of information? Because it's only as accurate as as we give the information? Correct? Correct. Pete Daddio: There's a lot of partnerships that are happening between electronic medical records, and companies like Microsoft and AWS, that they're partnering to be able to broker that conversation, in the sense of using a third party technology, or an application to be able to to either do the translation or have the platform understanding of this healthcare data. How can it be translated in a way, whether it's dictionaries, or having information that is brought in to be able to help to do those translations in another step too, and especially at a cancer center, like the one that I work at? patient records are just enormous, because when we get a patient that comes to Moffitt, they're usually coming with a whole notebook or a whole bundle, a lot of historic data, they've gone to the general practitioner, they may go to specialists, and then they I hate to say, end up coming to a cancer center, because they've gone pretty deep into their treatment and in their, in their medical plan. So how taking that information also, and being able to put that into a medical record that we have, we're starting to look at AI for those opportunities as well. So that can help bridge that gap to your point to say, if we're saying certain keywords are not me, but if doctors are saying certain keywords, there's already a dictionary of information there or there could be what's called a heat map of that type of information that's already in the patient record to help either emphasize or enforce it.
Chyrill Sandrini: It's very interesting, right? And is the is the possibility that the system itself will start to learn and generate its own knowledge as it's being fed that information? I mean, we've all seen the movies where the computer gets smarter, and starts doing its own diagnosis of the patient. Is there anything that we could be worried about there? Pete Daddio: Certainly, and I think that's where those boundaries have to be put in place to say, if we have AI and has all of this data to work with, can it come up with diagnosis and diagnosis ease, if you will? I'm sure it can. I'm not a medical professional. But I'm sure there's enough data to say if all of these data points are there, it can determine a diagnosis. The key is, and I've talked to many colleagues, especially from my side, being a technologist of colleagues and on the medical and nursing side to say this helps enforce patient plans, if you will, and patient diagnosis. What we're
doing at Moffitt is actually just to pivot slightly, we're actually using a lot of imaging data to help with cancer treatment options, in the sense that we're putting in all this data, a lot of historic images and a lot of historic data. And it's running models to say let's say we've got 10,000 patient records, and all of these patients had a particular type of cancer. Let's feed All that data and have aI run models to say, here are outcomes, here are possible markers, here are things to look for, here's historical, genetic information that can help output. Because we are also a full research
facility as well. So it's helped augmenting our clinical practice. But with that said, it is helping our researchers receive these results from Ai, to make better decisions to say, You know what, out of these 10,000 patients who have this particular cancer, we can now predict patterns, to help with treatments to help find the cancer earlier, and things like that to really help our medical professional. So my job as a tech, as a technologist is how can I help bring technology to run these tests, to do these things, to run these exams to be able to feed all this data to make sure all this data is there, that these tests can be run? Chyrill Sandrini: Right? Yeah, that's really interesting. I think, if we're able to share that data across the US, across the nation across the world, a lot of people could benefit from that, right. Pete Daddio: Absolutely. And that is actually one of the
outputs that we're looking to achieve is let's take all of our data, and be able to model it, and then have a monetization opportunity and a shareable opportunity. But first, we have to make sure it's de identified. And then so we're actually working on an AI process to feed in all this patient information, then to be able to de identify, because then that makes it shareable to be able to use your research. So let's D identify the data, then lets you use AI, again, to make sure that data is de identified. And then once it's the identified, it can be put into we'll call it a data pool that can be shared amongst other organizations, other research organizations, other healthcare organizations, to help drive better research better outcomes. And to your point, absolutely sharing that data, and hopefully finding cures for cancer.
Chyrill Sandrini: Yeah, what? That'd be awesome. Ah, that'd be amazing. Pete, what about this smart room, you kind of piqued my curiosity, because I really think the patient experience helps healing. How a patient feels in their room helps them get better, quicker, recover quicker. Tell me about that. Pete Daddio: Sure. So we actually just opened up a new
hospital about five weeks ago. And in this new hospital, we have to your point smart rooms. So the smart rooms have a lot of technology in them to help bring a better patient experience. We have Amazon echoes. So patients and our family members can ask, I'll say Alexa hopefully doesn't make anybody's Alexa go off. But
as to be able to add to be able to ask questions, and to be able to do things like maybe order food or change the channels on the TV. We also have virtual whiteboards So, and typical, or I don't want to say old fashioned patient rooms. There's a whiteboard at the foot wall of the patient, and doctors and nurses will come in and ascribe notes, who is your nurse today? Are there any foul risks and things like that? Well, we now have a digital display to to show all this information. And we also have, we use a system called RTLs for tracking so when a nurse or a doctor walks into the room, there they have a badge on it will automatically display their name to say this is, who your nurses are, who your doctor is, and what the goal of it is, is with other things as well as to typically in a hospital room, it's always been a cold, in really just kind of dire situation, we're trying to make it more make it feel like more of a home, you know, and it's not just for the patient as well. We have also brought in technology, like
tablets to help with caregivers. We also have a very nice couch that has a bed that allows caregivers who will probably be spending quite a bit of time in those rooms with their family member with their friend, and be able to give them entertainment options and for them to be able to relax and enjoy as much as they can, especially in a situation where somebody's going through some level of cancer treatment. So we've enabled these technologies and we're actually looking at how can we do more? We're looking at virtual visits to where let's say we have family members who are not local, they're crossed the country. Well, let's use camera and have them come in and
do a quick video but have a quick note. They don't need to go through a number of steps just have it readily available. Chyrill Sandrini: Yeah, that'd be awesome. You know, I lost both my parents to cancer and being a caregiver During those those patient times in my mom's room or dad's room, and even at home, there was so much information, like I had to pass along or regurgitate. And maybe how they were feeling what they were feeling. Do you see that AI can also help that experience that would benefit the clinician on knowing maybe what's going on at home or, or what the caregivers experiencing with the patient? Pete Daddio: Absolutely. And I think the key is to make sure
all the data is available. But to your point, it's how do you display it or provide it to the caregiver or the family member, or the patient themselves and not have data overload. And I think that's one thing that we're working on. Getting me as part of that patient experience is using the technology we have in the room, to be able to provide the patient and and or caregiver, the information that they need, making sure all the data is available. However, this is just what they're looking for, whether it's what's their treatment plan, what is their after care visit, here, the things that you need to do when you get discharged. But doing it in a way that does not overload the patient. And then be able to have a continuum of care with
what we're doing is called Digital front door, where then a patient can access our portal, outside of the hospital outside of the clinical area, to continue that care, here's your discharge information, here's your six month follow up and have all the information but again, not have it in a way that's at their fingertips, but it's also not overwhelmed. So and also working on that to help the patient but also the patient provide that feedback and make it better each time. Chyrill Sandrini: We know I really like that. You feel pretty alone, sometimes you kind of forget information, we kept a notebook. And I feel like this part of technology is really going to help streamline that. And we hear different, right you
and I can be in the same room with the same doctor. And we hear choke totally two different things. So I think that would be helpful as well. And you know, we Google everything, right? So having this information more readily available from a clinicians perspective directly relating to that patient, I think it'd be more beneficial. Pete Daddio: Definitely. And I think with the technology that
we're providing, to be able to build that out efficiently to really have particular overviews per patient, if you will, you know, for per patient experience. Excuse me. This can help because to your point, some can really absorb the information well, in some cannot, some may be coming in as a cancer patient, just completely thinking of so many different things. And unfortunately, because they're going through maybe one of the most traumatic things will ever go through in their life. But how do we make it where the
information is available to give that patient and or caregiver the opportunity to digest it when they want and how they want, you know, maybe I just need a summary. And I know there are companies that are working with our medical records platforms to be able to provide augment and vote opportunities for that to say, I just need a one liner of what my expectations are when it comes to my medical diagnosis, or, you know, what gives me the full complete overview of what my plan is, and then have aI potentially help with that. And Dr. To say, you know, what's real? Here's what works for you,
Pete, here's what works for you, and have it be more personalized. Chyrill Sandrini: That's great. So how long did it take at your center? I know I've heard been asking a lot of questions out there. You know, people are using it like you say an imaging and people are tiptoeing into the water, so to speak. How long
does it take your center to trust the data your clinicians to trust? How what was the timeframe? Like? Did you walk in the door? Was it you bring it in? Tell me a little bit about background and where you're at now? Pete Daddio: I think we're still going through some proof of concepts, if you will, because and I think it's when it comes to the trust, I think it's the amount of data that needs to be input, if you will, and again, being a cancer specialty organization, the amount of data we have is tremendously high per patient. So while you can go certainly to an ambulatory center, and there's plenty of data there as well not to to minimize that. But I think for cancer patients because of the totality of data. We're trying to find where that value is, is
it when it's an initial diagnosis where patient comes in the first time or is it through treatment plans and taking all that data that comes through treatments follow up And then the continuation of that, because there is a history now, I think we're trying to figure out where the sweet spot is, if you will. And based on some of the feedback I've received, I think that's where we're trying to go is where is that against sweet spot? But it really is, what's the availability of the data? And because of the large amount of it, we're still trying to get it loaded into the system, if you will. So we're making strides, but it's probably going to take some time.
Chyrill Sandrini: Yeah, I was just thinking, while you were talking about that, is pharmaceuticals, the pharmacist at the hospital, always trying to create that right cocktail, maybe for infection, you know, and cancer patients, you know, deal a lot with that pharmacist, or they're very specialized, I've seen it firsthand what they can create to combat a lot of things, including pseudomonas, let's say, and it's a little bit different in every case, and we've got to find out what works. Do you see the clinical and the pharmaceutical aspect, coming together using AI, sharing that information, Pete Daddio: I would say there's a definitive possibility. And I think the big driver for that will be the data to say, here's all this data of patient history and patient records. And to be able to say, if we've got a patient who's got these either symptoms, or a patient who has this history, the AI generation will be able to to to do predictive capabilities to say, this type of either drug, or this type of outcome is probably the best for this case. And I think that's going to help a lot of our researchers as we move forward. So I think it's really
getting a lot of that historical data, and using AI to be able to model appropriately for pharmacists to quote unquote, try different things. And I think that's certainly going to help in the cancer space with some of the clinical trials that we see. And to your point, coming up with different drug cocktails to maybe help combat cancers, if you will. So I definitely see that helping driving towards better outcomes in the future. Yeah, Chyrill Sandrini: I can see that as a big benefit, especially when we start to share that information with one another other facilities across the United States. Now, that
technology infrastructure, where did you start building it? And how is it now? And what how is it driving your AI process? Pete Daddio: So one of the challenges with healthcare is we're not necessarily up to speed or you know, ahead of the curve when it comes to technology. So with that, I think it's bolstering the opportunity for us to say we've got legacy information legacy systems. And it's really forcing us as technologists to find new ways to keep technology up to date. So with that being said, we're using more of cloud opportunities and using more cloud applications. And we're
starting to put more data in the cloud because it makes it a little bit more robust. shareable, as we were saying with the potential for shareable data. And we're using we're starting to look at AI to be able to augment just some of our technology processes. We've got to update our systems in our servers. How do we automate that with AI. So even though we're talking about very robust clinical and medical operating opportunities, we're looking at AI that just do some keep the lights on activity from a technology perspective. I'll
give you a quick example that I had one of our administrators, who's on one of my teams, he was trying to write a script to build a report. Because then I asked him, he said, You know what, I'm gonna go in the chat GPT and have chat GPT provide the script, certainly provide it within five seconds, copy the script into his program, and 30 seconds later, you had a report. So it's those types of little scenarios that are also we're finding the benefit of AI. And it's building those small steps into the bigger ones to help with technology to help drive those things moving forward. So I think it's elevating our maturity through AI to get kind of past the challenges we've had with trying to update technology, if you will.
Chyrill Sandrini: Yeah, somebody asks a million dollar question for our listeners, how does aI have set? The biomed be met? How is this gonna affect them? And do you see that integration in the future? Pete Daddio: I certainly think there's the potential for it. And I think the biggest challenge is going to be the security and ethics around it. But I think there will definitely be opportunities for the biomed space and technology space to be able to work to help us that type of equipment, use those types of programs to be able to have better patient outcomes. And I think it may, it may continue on a smaller scale
to bring bigger results instead of going for, you know, the big one, then having all the issues of well, they're security questions and things. But let's start on smaller scales and build from there. So and I think if we can find the right use cases to move forward, it will definitely have a positive effect. Chyrill Sandrini: Yeah, I agree. I mean, do you trust AI.
Pete Daddio: So my motto is trust the Verify. And I use AI for certain things. But I don't use AI to make decisions. I use AI to help me make decisions. So I look for AI to help augment either activities or processes that are time consuming, to help me with better outcomes. So for me, it's Trust, but verify. I
think AI is a great tool, I think it's a great opportunity to be able to emphasize some things that we're doing. But we still need the human contact and the human part of it to be able to make the right decisions. Chyrill Sandrini: So would that be your word of wisdom for this episode? Pete Daddio: Trust but verify. I can't say that enough. Absolutely.
Chyrill Sandrini: I mean, it that's, that's so true. But isn't it true of everything in life? Pete Daddio: Absolutely. And I can give you one quick example. That is in, not in in the healthcare space. But there was a story back in June, where there was a law firm that used chat GPT, to build out their closing argument for a case. And I don't remember the particulars of the case. But come to find out that the data they used to build out that closing argument was actually bad data. So when they went to do their closing
argument, it was sound, but it didn't work because they were referencing cases that didn't exist. So I always use that as my example, to say trust but verify, use the tools. Absolutely. There's so many ways to use AI these days. But we
have to build that trust. And we just have to verify what the outcomes are. Chyrill Sandrini: You've heard it here today, listen, listeners with AI Trust, but verify. That's the takeaway from this.
And I think it's so exciting, Pete, and I really appreciate you coming on. Because I know this is such a popular subject on many platforms, not just healthcare. But what could it do for cancer research? What could it do in the health care space, if people aren't as sick for as long? Or maybe we get to a place where we prevent it. Maybe we can prevent cancer? That would be amazing. We can cure it. That'd be amazing. To me, people suffer out there. And so I think AI is a great opportunity for all of us to learn from but as Pete said today, trust but verify. Thanks so much, Pete for coming on. Thank you, Shirley. I
Pete Daddio: really appreciate the time. Chyrill Sandrini: I hope we get to meet in person some date down in Tampa. And hopefully you have no more hurricanes in your future. Pete Daddio: I hope not. Chyrill Sandrini: Yes, I've heard you know, I have some friends in that area and some of them got hit hard some spots are missed and you know, we're all sending you positive thoughts down it down to there.
Pete Daddio: And you're always welcome to sunny balmy Tampa. Chyrill Sandrini: I will be there I will be there soon. I promise. So you guys follow us anyplace we listen to your podcast you know we're on every every place you can find this Spotify. I think it's called the Apple Music now and not iTunes
anymore YouTube. You can always find all of our episodes hosted on our website at MultiMedical systems.com. If you're interested in becoming a guest or you have a topic you'd like us to cover, please reach out to me again, Pete. Thanks for coming on. And we look forward to seeing everyone next time on
htm insider
2023-11-20