Welcome to consensus, a podcast from Census Technologies. Hey. What's going on, y'all? It's Daniel Littwin, the voice of b two b, and welcome to another episode of the consensus podcast, a Census Technologies podcast. It's good to be back in the Census hot seat here. This time, we're sitting down with two guests from within the Census ecosystem to talk shop on a big trend. You know, a buzzword, obviously, everyone's talking about AI and the role AI is gonna play in different departments, different fields, different verticals. Well, we're gonna have our take at a little AI conversation, but we're gonna cut through the noise. We're gonna get practical, and we're gonna be discussing how AI can, should, and already is being used in sterile processing departments. And we have two great voices here to shed some light on those best practices. So I'm joined today by two leaders who bring both the technical vision as well as the domain expertise to our conversation. I'd like to go ahead and welcome the two of them. Welcome, guys. It's awesome to have you guys here. Let's introduce both of them. First up, it's Harshil Garradia. He is CTO, chief technology officer at Census. He helps shape how emerging technologies like AI are thoughtfully applied to health care workflows. Harshil, welcome. How are doing? Hey. Great. Great to be here. Thank you so much, Daniel, and looking forward, you know, to this episode, I think we we are really at the cusp of, you know, driving a lot of impact with AI for our customers. And, you know, we we are super excited to be in this journey. Thank you so much for being here. Absolutely. And welcome to the show. We're also joined today by mister Seamus Johnson. He's senior director of innovation at Census, and he brings deep sterile processing knowledge and works very closely with customers to turn real world challenges into practical solutions. Seamus, welcome. How you doing? I'm doing great. Thanks for having me. This is this is very exciting. Yeah, as you said, I've been at Census for about twenty years now working with customers to understand their problems and trying to figure out is there something that we here at Census could and should build to help solve those problems? So, you know, with the rise of AI over the last couple of years, this is a very exciting time in this industry. I'm super excited to be here. Yeah, and we're excited to have you. So let's jump in folks. Again, we're going be talking about AI and sterile processing, in workflows, in departments at large for specific tasks and steps in SPD workflows. But we're gonna also give a little special focus here on Assembly Co Pilot final check, right? Which is a real example of how AI is already helping SPD teams reduce errors, build confidence, strengthen patient safety. Right? So I'll be looking for some anecdotes from you guys, maybe even some hard data to help guide our conversation. So let's get into it. I wanna start with Harshal from the CTO perspective. Right? You're overseeing the big picture here for Census and how y'all solutions and services help to support the big picture changes that are shaping health care today. So from a technology leadership perspective, why would you say right now is the right moment for AI and sterile processing, if at all? Right? What are your takes there on just kind of where we're at here in this moment of AI adoption across health care and in SPD? Yeah. So I think this is the right moment really for using AI in sterile processing. I think primarily because of four critical trends have converged, which is making AI both necessary and practical in SPD. First is the workforce staffing challenges which we see in SPD. It's real. SPDs are facing chronic persistent staffing challenges. And it's actually being compounded by the fact that the case volumes have increased, complexity have increased. We have seen a lot of upticks since the pandemic. And unfortunately, SPD staffing has not kept pace with that growing volumes. And in fact, you know, was just reading an article about five percent of patients actually experience surgical site infections just because of improper sterilization, right, which is directly underpinning, underscoring how SPD performance affects these outcomes. So AI can be a big enabler here. The other part here, second one is data. Right? SPD teams, systems rather, now generate rich structured data. We do instrument tracking. Millions of assembly logs are captured, OR schedules, sterilizer cycle records. So that data is really creating that fuel for AI to detect patterns, predict risks, and provide those reliable decision support for SPDs. I think the third area I would touch upon is AI itself has matured. You talked about that, Daniel, you know, we are, you know, really moving forward from an experimental stage to actual deployments. Modern AI integrates with workflows. It provides assistive intelligence. So it augments the human judgment, right? So we are already seeing AI assisted visual identification of defects, tray completeness, missing instruments. So that can be a big enabler, right? So AI technology itself has matured. And the last one I would touch upon is, you know, the regular regulatory compliance requirements, right? So we have organizations like AME, regulators like the JCODE, the Joint Commission, which really keep hospitals accountable for sterilization practices. And AI strengthens the SPD's ability to meet all those expectations. So all in all, I think AI can be treated as a real safety partner for the SPDs. Now Seamus, let me toss it over your way. From the product and the SPD side, right, where you've really planted your roots, you have a lot of hands on experience, give us the team perspective here that's guiding AI implementation in the first place. What kind of pressures are teams facing today in sterile processing departments that you would say make this AI conversation timely and relevant? Give us that voice from the field. I think SPDs today are really sort of facing a perfect storm. There's a couple of long term trends that have gotten worse over the last twenty years, and they're all coming together to really put a lot of strain on those departments, which in turn puts patients at risk. When I started here in two thousand and three, if you worked in sterile processing, you could make twelve dollars to fifteen dollars an hour, which back in those days was a step up from most starting entry level positions. Today, sterile processing doesn't pay much more than that. Dollars seventeen, eighteen an hour is pretty common. And just downstairs from our offices there's a restaurant called Radish that has a sandwich board out front saying they're hiring at twenty dollars an hour. That makes it really hard for the hospital across the street to staff sterile processing. You know, if you think about it, if you're somebody right out of high school, why would you go work in a, you know, a hot, wet decontamination environment where you could get stabbed with a K wire and get a really bad disease versus, you know, working behind the counter with air conditioning, making more money, right? When the pandemic hit in twenty twenty, that was a big inflection point as well. You know, a lot of hospitals stopped doing, elective surgeries for a while, ended up furloughing people in sterile processing, and many of those people left the industry and never came back. So there's an economic challenge there around staffing. There's also, and Harshal touched on this a little bit too, there is this long trend, driving complexity in the cases. Know, it wasn't nearly as common for robots to be doing surgery. Today, that's pretty commonplace. Every hospital I know is buying a surgery robot or already has one. And so processing instruments has gone from, you know, in the 1950s and 60s, washing a lot of stainless steel parts, the equivalent of dishwashing, to today being a very technical and very demanding job. It takes a certain level of training and aptitude to do it and to do it well. And with that, you've got really the third challenge, the third part of that perfect storm. And that is the education around sterile processing, the training has not kept pace. If you talk to anybody who works in sterile processing about the challenges around IFUs, you know, it's very real. There are hundreds of thousands of different types of instruments, and that's not an exaggeration. The manufacturers provide IFUs, steps for how you have to actually clean them. But those steps are put together by lawyers. They are designed to protect the manufacturer, not to educate the technician. And so they're really setting up sterile processing for failure. So they're understaffed, they're under treated, they're under supported, and that has a real impact on patients today. Well, there you go, folks. I mean, it sounds like there's no better time for us to be thinking through, I mean, really every layer of that operation. It's not like AI is one silver bullet that's gonna solve for everything you just broke down there. Because there are a lot of macro industry level trends there that are placing pressure on the industry. Network to network, hospital to hospital, you're gonna have nuances and some of these dynamics, but you're surfacing some key themes there, right? Increasing case complexity, staffing shortages, burnout, having to do more with less, rising expectations for accuracy, documentation, patient safety. So it's like at the time where it's a little less desirable to go start your career in an SPD, the demands for SPD just keeps going up and up, right? So it seems like the perfect opportunity for a layer of technology of data driven insights powered by AI to help triage, pun intended, I guess, right, some of those issues here. So I wanna make sure you're heading to our website. So make sure you guys are heading to census dot com. Again, that's c e n s I s dot com. There you'll find more information on our solutions and services, how we support several processing departments, but, of course, more episodes of the consensus podcast as well. You can also subscribe to the consensus podcast on Apple Podcasts and Spotify. So just hit that subscribe button. You'll get the full catalog of previous episodes. You'll get notifications when we drop new thought leadership, both conversations with our own Census team as well as with our wider community of SPD pros. So I wanna just open up questions here for the two of y'all now to just talk a little bit more about the problem that AI is really solving in SPD and where we need to be thinking as SPD strategists for just like where to deploy AI in the first place. So I'll open this up to you guys. Where do you guys see the most risk or inconsistency in SPD workflows today? Yeah, I think it's really in being consistent. Like, so if, you know, the places where things can go wrong are the same places today as they were twenty years ago. It's decon, it's assembly, it's sterilization. Doing those correctly, doing them, you know, according to the IFU, and doing it the right way every time is the real challenge. You know, technicians are wonderful, wonderful people, but they're people. And people will eventually make mistakes. So, we'll get to it at the end, but that was really what drove our thinking behind the Co Pilot final check solution. You know, at any given hospital, we run stats on this, they process tens of thousands of instruments every month. It's not out of the ordinary to see a hospital process thirty thousand instruments in a month. That's a lot of instruments and every single one of them has to be cleaned properly. Every single one of them is a little bit different. And so, you know, helping the technicians to do that consistently, that's the challenge, I think. And, you know, the staffing that we're talking about, staffing challenges exacerbate that. When you're understaffed and you're pushed to go faster that makes that so much harder. Harshil, any follow-up thoughts on that? Yeah I think Seamus touched upon that right I think the consistency is the key, the predictability is the key, right that's what OR looks at it. SPD has a very strong relationship which they need to continue to build with the OR, right? If you just think from a hospital ecosystem, the OR is the one which drives your cases, case volumes, right? That's where hospitals you know apart from the importance to patient safety that's where you know the reputation of the hospital is at stake you know the revenue of the hospital is driven based on the OR volumes and how can SPD be a big support player in this ecosystem is absolutely critical right so some of the things which Seamus highlighted upon being predictable dry have the right instruments at the right place at the right time one hundred percent of the time cleaned and ready to go is what we are trying to solve here. So how do we get to that zero defect predictable outcomes one hundred percent of the time? And that's where I think we think AI can play a huge role, right? It can actually support through automation, value added automation, you know, make lives easier for the SPD to achieve their goals to support the OR. So let's pull at the tool thread here a little bit more. I'd love to hear from both of y'all on this, though. We can start with Harshil. From an AI standpoint, why are these types of problems difficult to solve with some of the traditional software solutions that are already out there for sterile processing departments? Right? Like, why why is this an AI conversation and not just a general tool or software conversation? That's a great question. And and, you know, we we think about this all the time. Right? Like, you know, there are so many tools out there, you know, there is always that decision about, you know, do we need to invest more in a new AI based solution versus, you know, can we leverage what the customers already have? And, I think the big thing which we have seen here is the traditional software is great when it's rule based. What I mean by that is when you can build those rules in the environment, which can evaluate and make decisions on all scenarios, right, where it knows that, you know, if it's scenario A, this is how it should behave. So to give an example, right, if you're doing a tray verification, if you say, you know, I need to, there is instrument A in this tray for an ortho tray. If that's, if you have that instrument A, you need to have all these other fifteen instruments placed in this way. Alright? That's a very predictable scenario. It breaks down when in a in an SPD environment because a lot of these data points are visual and contextual, right? So to take this further, right, in that same example I gave, what if your instruments are moved in a different position? What if there is light glare coming in on that tray, right? What if they are stacked differently, right? So you can't really build rules for every such complexity and every such scenario, right? Because it's very contextual, it's visual and that's where, you know, AI can play a big role because it can interpret these images, it can learn from variability, right? So you're not hard coding those rules and that way it can surface those risks before they happen. So I think that's where AI can play a big role here, right? It's a very, very valuable That's what makes this really exciting too. Know, I said at the beginning, this is an exciting time to be in this industry. AI gives us an opportunity to solve problems we couldn't solve before with traditional software. So just like Harshal was saying, when we look at the problems that we're trying to solve today with AI, they are the same problems we tried to solve for twenty years with traditional software. And some of those are very difficult. We'll get to it when we get to final check, but we've seen customers and ourselves attempt to use traditional software to do things like prompt reminders for technicians to take an action or pop up a message to give warnings about certain pieces of information. That's very easy to do with traditional software, but like Harshal said, you know, interpreting complex real world data, seeing the world and understanding what's going on, understanding it the way a human would understand it, that's not something traditional software can do. And, you know, ten, fifteen years ago, if you told me this world was going to exist, my head would have exploded. So now we have these opportunities to go back and solve these, really important problems, these persistent problems and elevate SPD in the process. Yeah, you can always just, you know, just to add, to conclude on this, right, you can always think about traditional software being more reactive, and AI can be proactive. That's what we need to drive all the critical outcomes and improve patient safety. So let's dig in a little bit more then into Assembly Co Pilot briefly here. This is really a critical breakdown, you know, not to self plug too much, but Assembly Copilot final check is that AI layer designed by Census here from the ground up with these nuances in mind, with the need for proactivity around SPD challenges, with the need for supplementing stretched, maybe understaffed, under resourced, sterile processing departments, with those workflows in mind, especially knowing which areas of those workflows are very difficult to improve upon manually, that require more insight, insight that goes beyond any one person's day to day experience, you know, and using that anecdotally to polish up an entire operation or workflow. So let's dig in a little bit more into how this was designed. So Seamus, can you give us some insight here as our innovation lead? What specific customer challenges? I mean, I know we've already riffed on some of them, but anecdotally, if anything comes to mind, what are some of the specific customer challenges that led to the development of Final Check as a supplemental AI solution? Yeah. So first off, I wanna give a huge shout out to our customers. You said this is something that Sense has built, but we didn't build it in a vacuum. We built it in partnership with our customers. We got a lot of feedback from a lot of customers. Huge shout out to them for helping us understand those problems and then work through the various solutions to find something that actually worked. Yeah, so we, you know, for as long as I've been here, there has been a challenge around consistently putting integrators into trays. And it's not just putting integrators into trays, there's all kinds of issues with trays. This is the one that we decided to tackle first. You know, was talking to a technician up in Ohio, and at that facility, they had created a reminder barcode that the technician was supposed to scan after they put the integrator in the tray. Just something, go through the motions so that you don't forget to do it, right? You have to do this step. And I asked her, said, how come even with this reminder people still forget? And she said, you know, we get into the zone. We're in the flow. You know, when you're working eight hours, she said, I like to stay busy. And so I, it's all muscle memory. I do this, I do this, I do this, I do this, I click here, boom, done, next tray, boom, boom, boom, boom, done, next tray, right? And you just kind of get in the flow and you zone out, you don't think about it and you make a mistake. And, and so we're trying to figure out, well, how do you solve that? With traditional software, you really can't. And the idea was what if SensiTrack can be a partner for the SPD technician? We're not ever going to replace SPD technicians, but we can be a partner for that SPD technician. The computers are very good at consistency, doing the same thing every time. And so that's really where the idea came from. AI is the key to unlock all this. That's not something we could have built fifteen years ago. But today with machine learning, with the ability to write computer programs that think and behave more like humans, that can interpret complex data, like an image of a tray, and search it for that integrator and see if it was placed in there properly, that's something we can do today. And, yeah, I wanna pull at this thread with you, Seamus, for a second of the back and forth with customers here. Can you can you expand a little bit more on how Census users help influence y'all's decision making on new features, new solutions. How do you bring your community to the table to learn from them and better understand what they need as you're in the design process of new tools? Yeah, it's a lot of talking to customers, going to their facilities, visiting the customers, seeing things in the real world. There's absolutely no substitute for that. So, I think there are a group of customers that I talk to every week. I talk to them individually. If anybody out there is listening and you'd like to talk to me on a regular basis, feel free to reach out. I love talking to customers. It is honestly the highlight of my day every day. And so it's an iterative approach. Talking to customers, getting them to look at different ideas we have and give honest feedback. You know, I want people to channel their inner Simon Cowell and tell me when my baby's ugly early and often, right? That's absolutely true. I don't want to go build something that nobody's going to like. So, you know, we put hundreds of ideas in front of customers over many weeks, shoot holes in things, refine things, and eventually come up with a design that everybody believes in. We go build prototypes, we go to their facilities, we test those prototypes with the customers, we watch people use prototypes to understand how the users think about it, how they discover features. We look at performance metrics, we look at, you know, anything that might frustrate a user. We look at all kinds of things, but it is really all in partnership with our customers to build the final product. And I just like to add here, you know, I think it's a very structured process. We follow your, you know, census as part of Fortive and we follow what is called as photo business system. You know, it's really ingrained into our culture of, you know, how we design new products, right? These are things, you know, we, as Seamus said, work closely with our customers, get their feedback, we are building products, we are deploying it, you know and we are improving it together and that's a very very structured process. I mean if we have seen you know that as if customers actually get their opportunity to give real insights and be part of that whole design process, the final outcome is much much better than what it would have been had we just relied on our internal experience, Right? And and that's really ingrained into our culture of FPS. So you you expanded on some of these, Seamus. Right? You know, AI being a final check to ensure that there isn't rework, that there aren't trade delays, right? Having AI really be there to watch out, to make sure that these small mistakes don't turn into a catastrophic failure that then impacts the OR. Another layer of this I wanted to get your quick thoughts on is just the the need for consistency across technicians and across shifts. Right? In previous conversations I've had here on the consensus podcast, this theme of a positive team culture in SPDs keeps coming back into the fore. How important it is, how everyone being on the same page of not only how do we do our job well, but why we're doing our job helps to enforce these habits and cultures of attention to detail and, you know, being able to kinda muscle through some of the challenges in today's SPDs for the sake of the role you're playing and and really being like a last line of defense for patient health. Now, obviously, building that kind of consistency across your team can be difficult when you are a large team, when the lesson you learn at eight AM is not the same lesson that they do it at eight PM learns. You know? So having these tools that help almost knowledge share, right, and and proactively reinforce those standards, I imagine benefits and is benefited by, you know, the same kind of culture conversation. So I just wanted to get your thoughts there too on how you see a tool like Final Check helping to connect those dots across shifts, across technicians and build more consistency. Culture build in the hospital, particularly that relationship, and Harshal talked about this earlier, between the OR and sterile processing is very critical. The best customers that we have have created a collaborative relationship and it's built on trust. Every now and again, I see a hospital that is in distress. And I know everybody's either lived it or heard about it where there is a very caustic, corrosive relationship between the OR and the SB. And that's where you hear the worst of the worst stories. Trust is so important to that relationship. If you think about it from the nurse's perspective, they rely on SPD to deliver for them. When they can't trust SPD, that's when fear takes hold, and that's when you hear about the worst of the worst things happening. What final check does is it gives sterile processing a tool to help build that trust. It helps ensure that they're delivering trays with those integrators every time. In all of our beta tests, we saw customers with dozens of trays a month going up to the OR missing integrators. After they turned on the feature, they went to zero. Zero. That's unheard of in this industry. I've never seen another solution have that drastic an impact. And that's where you need to get to in order to build back that trust. The nurses need to know that they can depend on sterile processing. That is the foundation of that relationship. And when the two departments work collaboratively, that's when magic happens. Hey, what's going on y'all? We're cutting the episode a little short here today. We've got so many good insights. We want to make sure you're paying attention for the entire time. So we're gonna carve this up into a part one and part two. So that concludes part one of our conversation here with Seamus and Harshil. Make sure you stay tuned as we drop part two and continue to explore more use cases for final check the larger consensus solutions ecosystem as we drill down into more impact stories, use cases for AI in SPDs, and looking ahead to the future of AI in SPDs at large and how SPDs should start to strategize around some of those changes. So stay tuned for that episode. Till then, I'm Daniel Litwin, voice of b two b. We'll catch you on part two of this episode of the consensus podcast.
