Kay - Good morning, good afternoon. Good evening. Welcome to the experience dialogue. In these interactions. We pick a Hot Topic. That doesn't have a straightforward answer. We then bring in speakers who have been there and seen this but approached it in many different ways. This is a space for healthy. Disagreements and discussions. But in a very respectful way, just by the nature of how we have conceived, this, you will see passionate wiser of opinions friends. Having a dialogue and thereby interrupting each other or finishing each. The sentences. Our mission is at the end of the dialogue. We want our audience to leave with valuable insights and approaches that you can try at your workplace and continue the discourse in our social media channels with that.

It's my pleasure to introduce the topic of today. Which is AI and how to use AI to drive service improvements. And for this, we have several Eight. She is aum global medical device leader. And then, what interested me about Anne's background, is she has done everything from, our organizational strategy program management sales marketing field operations and she is ranked huge field service operations within Philips and Baxter in many many, many capacities and has grown. The business has considerably And she has a Ph.D. in biomedical engineering from Washington University with that. She focuses on the patient. Experience is fabulous and it fits in this conversation today. We're and will be taking us through a framework of how to introduce data science to improve service operations. Including how to identify a proof-of-concept project. So how do you start with a Concept and how do you determine AI is the right for your organization for this? She will be sharing a practical example of, how she and her team used AI data-driven insights. To drive improvements in service processes at large medical device companies and teams with that welcome and welcome to the discussion.

Anne - Thank you. Okay. I'm excited to be here and speak with you and those who are Watching this experience dialogue today. Very excited to discuss this topic.

Kay - Yeah, I know this is the first LinkedIn life for you. So it's a good experience and we had a lot of interest from service leaders who are in medical devices and outside of the medical device, backgrounds, who are in the show watching today, which brings me to the first question, what led you to impart down the AI Pack.
Anne - Yeah. So you know, in the years that I've been managing service organizations but the one thing that you have a lot of within Services, data tons, and tons and tons of data so much that, you know, it leads you to wonder how best to mine. It how best to make conclusions from it and how do you, you know, improve, you know, over time and so what intrigued me about AI, you know, we had started. Are working with you through the R&D group, you know, looking at some futuristic kind of applications and reviewing log files and things like that, and as you and I were discussing with the team. Well, what can we do today? You know, so certain improvements have to be made over time. You have access to log files and increase the information that's available in them. But is there anything we can do today? That would practically help solve some of these issues. And that's where we started to discuss. Do you know what kind of information is available and then what can we do with it to help Drive improvements within These large organizations?

Kay - Yeah, and usually what we have seen companies, look at implementing IoT on edge devices, digital twins data, lakes, and all of that, all of it was important. But I remember the first conversation we were having and you were like, what can we do with the service circuit data? Because we have very, you know, details of and what more value can be reaped from the existing service record data so we can improve service operations. And that's how I believe we started down the service record data path.

Anne - Absolutely. Yeah. And I had in parallel but having a discussion within my organization that the Service Experts and the service leaders across the globe around you know some potential you know needs that they had that they saw. Right and so just to give an example in this, this is related to the project that we ended up working on together. Many medical devices require some sort of annual maintenance schedule, right? Whether it's an annual preventive, maintenance work by annual, you know, And this is a huge driver of overall service cost because it's a predictable service event that has to happen at a certain frequency and requires a field service engineer to go out and look at the equipment. Well, most companies are looking at, you know, what is the optimal p.m. schedule for any given piece of equipment? It's not always assessed when the product is originally designed, it is sort of assumed. Well we need to do something annually, right? And so with this, Particular device that we were working with there was a biannual p.m. that was pretty detailed and involved parts replacement and several things that were kind of required to keep the equipment up and running.However, there was also this annual on-the-off years kind of a p.m. light, you know. So it was an electrical safety check and some you know some basic things that India decided years and years and years ago were required to make sure that the equipment Moment was running and functional and so our regional leaders and our Service Experts were sorts of asking the question, what is this p.m. Am I doing? You know, is it reducing the incidence of corrective maintenance later? Excuse me. Sorry. Yeah. And there wasn't a good way for them to prove that so they had gone to R&d. And said can we look at this and Hardy said, well it's always been there.It's going to take a lot of time and money to assess that we're going to have to run devices and do testing. And, you know, let's just leave it for now. We've got other stuff we're working on and so it kept coming up and that's when we started our conversations around. What is it that we could do with the existing data? And we realized we do have these service records, right? We have a history of what has occurred on these devices now Now, the question was, how do you know if you took away a p.m. light, for instance, versus keeping it in? You know, what would happen to the devices? And that's where we had a bit of a fortuitous occurrence that had occurred as well. This is that just like within all large medical device companies, occasionally there are some misinterpretations or inaccurate interpretations of requirements which had You know, within this device in a certain country for many years and that's a compliance risk. But you know we run into it all the time within service that these things happen and you have to look at it and then make a decision. How do you remedy that? But in this case, we had a country that had not been doing the p.m. light events for some time, and then we had the rest of the world where they had. And so basically we had a test Population. And then, you know, the control that we could look at the service records and compare amongst them and say.So in these groups where that wasn't done what happened were there more Service events that were required. The problem was. And so we knew this but the problem is the amount of data, right? I mean, we're talking thousands, thousands of service records, you had to track it over time from one device to the other. There was also manual information entered, right? So sometimes when you do a PSA, You know, they have to replace something that wasn't part of that p.m. schedule and we needed to document all of that. And so that's exactly the kind of thing that AI is designed to help solve, which would require a massive amount of people to do a project like that.And so that's where, you know, your team and our team partnered up and spent probably good two to three months, right? Figuring out how best we Analyze that information. And what are we looking for? And what do we do with the information that comes out of it? So, that's, that's kind of just in a nutshell, the project that we embarked on and we can talk a bit more about, you know, the results if you'd like. Or if there are more details around that, if you think

that the audience might like to know, I'd have to be happy to answer questions, too.

Kay - Yeah. You know, what intrigued me is the path on which Your team embarked on it, right? So there was a hypothesis. Hey, we needed to evaluate it. And you question some of the assumptions that have been there for a long period. That's the one-second thing you want to make sure that there is data, so substantiate, whatever hypothesis that we come up with, and that has to be cost-effective from a service angle optimized and quality based does. Snot impact patient experience and has to make sense of the data has to make sense, such that going forward. The teams can operate with this new normal and that's a perfect you know, a data-oriented project that you are describing here, you know to be able to do something like that with existing data and two to three months is awesome. And we also mentioned a little bit about humans.Into data and humans enter data. I should say so which means there will be some level of algorithm inconsistencies and any should factor into that level of data cleansing to see which ones to take and which ones to ignore and where to put the emphasis on and all of that.

Anne - Yeah. And it really required close collaboration between your team and then our team, right? Because you have to understand the process and the equipment and what's needed to be able to tell, you know, if we're seeing something a year later was that related to the fact that you know, the p.m. wasn't done or was it completely unrelated, right? And so the data can help tease that out but we wanted to look at that in detail together as a group because, you know, your team is the expert in the data. I didn't how to how to, how to develop those models and interpret the information that comes out of them and then you need the company to be the expert on the equipment right? To help point you in the right direction.
Kay - Yeah, absolutely. And that also means doing a lot of change management internally across the geographies with the product teams. Getting a lot of product feedback from service data and then driving product Efficiency through service data is not normal in this industry and you have spearheaded some of it. And we see that with a lot of other Med device companies that we are working with, is this a trend that you see continuing? How did you embark on the change?

Anne - Service is a huge cost driver for medical device organizations. Ask the globe, right? So it's a requirement, it's absolutely necessary. It provides a lot of value for customers as well. And so there's your huge attention on how to capitalize on that value while also reducing the cost, right? And that's true for the companies themselves, but also customers, right? So there's a lot of customers that do self-service on these, on these pieces of equipment, and they want to be able to manage how to do that themselves. And so the Question is, can they do it at the level that you know, an organization like Baxter Phillips does? You know, has that kind of knowledge and that kind of expertise within their group?

Well you know the only way to do that is to provide them with the right data. The right tools. The right information to be able to service that equipment at the right level.And so I see organizations large medical device organizations are very interested in AI-based solutions for service because it is, it is a way to improve efficiency for them, as well as for customers, and to create more value over time as those insights and that information can be fed back to improve devices. Number one and also to help customers understand how to maintain their equipment better. So yeah. Yeah, please continue. Oh no, I was just going to say I mean, I think you know, AI is being used all over and every industry.

But in particular in medical device service, Isis is one of those areas that can benefit the most from AI because of the massive amount of data and information that is running through a large organization every single day.

Kay - We are so happy to have been popped off that Journey with you and your team. And what benefits do you see from doing this project?

Anne - Yeah so I think, you know what What comes to mind first and foremost is always returning on investment, and what is the fine? What is the potential financial benefit? Right? And so, if we look at the example that we talked about with, you know, potentially removing this annual p.m. light, because the data showed that it didn't really, it didn't reduce the number of Service events that were happening with these devices over time. That alone and that example would have saved the organization a million dollars. Just to and that's every year, right? That's an annuity over time. So that's the kind of you know, what are relatively small projects that you can work on that have humongous Roi overtime right now. It's not just about Finance though. Service is all about customer experience and so you have to be able to prove that whatever change you're making first of all it doesn't degrade the customer experience in any way or another quality and compliance. Those are huge. Those are always number one and then beyond that. Is there an improvement? You know, for instance, in this case, if we tell customers will you don't, you don't have to do these p.m. lights either. If you're going to do self-service, right? That saves them time and money. They don't have to pull that equipment out of service for out of use for a day to do this. So that's a huge benefit to them and you can imagine and that's just in this example, but say that you Had access to service record data beyond this, and could look at Trends and patterns and see that certain devices are failing more frequently than others. For some reason, then that becomes a huge indicator of a customer experience issue or you might, we might need to pull the device out and then replace it or, you know, somehow determine which devices are functioning at the right level and not. And you could also avoid field actions, and narrow the scope of a field action if you could figure it out. For instance, it's only devices that were manufactured in 2015, that all of a sudden have these issues, that's the kind of power that I can give it can give that kind of insight. And ultimately, you know, help all of the metrics that a service organization is tracking, as well as, you know, improving the financial side as well.

Kay, - You bring up a very good point. Most of the teams are used to looking. In meantime, between failure, and them tbf failure characteristics alone in service what you are bringing up is that alone is not enough, there is a lot more color to it which is when it was installed. What is geography? Who was the supplier?Do you know which teams worked on it? What are the human element and I can keep going? Can you think of some more failures? Chicks that you can add to what we were just talking about. Site of alone, MTBF

Anne - Customer use patterns, it could be a chemical thing as well. That the way that they're using the equipment is somehow different in one region or another.It could be that the service process is not ideal in one country or, you know, with one set of tools. I mean there are so many variables and unfortunately, the tools to be able to assess all of that have been Limited. Did you know, for organizations and so they've had to embark on very large projects, to look at that information to try to narrow down the scope of field action, or to figure out the root cause or put together a Kappa, you know?So, these are ways that, I could speed up all of those various categories,

Kay - each one is a problem in itself, right? So neat. On that. You're pointing out can drive service improvements significantly on its own.um, So can you expand a little bit on the other things that you mentioned here? You know you said it fast. You can. Please expand on it,

Anne - sure.So, you know, if we're looking at some devices that let's say, you know, a piece of equipment was designed for an average use of Three times per day by a customer. But you're not, you're installing devices in a large high volume. A dialysis clinic is an example and they're doing 15 procedures per day. So most of the time, you know, the requirements that a device was built under our, then how it is tested, right? So for a device that's on average used three times a day, there's a certain service interval and frequency, and there's an Acted failure rate of certain Key Parts which are all within the acceptable range of how the product was originally designed for the customer requirements. But when you install those devices then and start using them just like you would a car, you start driving it. A lot more we have to do to get new tires, more frequently of to get the oil changed more frequently, and up until now, most companies haven't had the tools to be able to assess what is it that we need to do to make sure that that equipment. Payment is still delivering at the level that that customer expects. Right? And, do we need to modify the customer's expectations? Yes, we told you. It's only going to need one p.m. a year, but you're using it five times more than we expected. And so we need to do p.m. at least two or three times a year. Those are the kinds of insights that I think AI could help provide because of its real-time use. So it's unrealistic to design a product to Encompass all of the In areas, in which, it might be utilized in the field, they try, but it's very difficult to do, right? And so, the question is, once it's out there and they have real information about what's happening, how do we utilize that to then feed back into our service processes? Our data, our design requirements for the future to improve and I think that's where you know the types of Tools that Ascendo developing and putting together around a, I could benefit organizations in that effort.

Kay - Yes, Anne what you did within the, you know, your team is, you're getting information from the product into service, but you close the loop going in from service, back into the product into the design of the next generation of the product, giving them insights. What you see in the field. So, in a way, you have alleviated, not just the service experience for the patients, and our customers, but you elevated the service experience for the R&D teams to absolutely.

Anne - And I think it also has, you know, one of the challenges that a med device companies run into is that that feedback loop isn't always there as you just mentioned but even if it is there and it's getting into the next product, you still have 10 or 15 years of use of the existing product that you have to figure out how to optimize and it's unrealistic to read design the products that are out there, right? If there's a very large installed base there there there, and we need to figure out what to do with the ones that are already out there. And again, I think that's where I can help. So that's, that's sustaining engineering piece is huge for a lot of companies. It's a Strain on R&D, resources, and investment, it's necessary. But if there's a way to sort of provide better data around, what are the top opportunities? Because I think it gets hard, there's a laundry list of items that, you know, as a service organization, you want to see improved in the devices that are out there but R&d and sustaining engineering rightfully ask well, which ones are we going to go after? Because I can't work on 50 things. And so that's where having better data. Helps. You build out a case for which ones can require sustaining engineering resources, which ones are a process issue that could be solved within the service itself, or which ones do we just literally need to replace the devices because they're just not functioning at the level that was intended, you know?
Kay - Yeah. And that's, you know, the beauty of using something like a simple AI is As you said, we can get that pretty much real-time information back from service as a voice of the customer into the product. And that gives a lot faster feedback back from the field into the product. And like you said, it's for sustaining, how do you manage and continue? The experience of the existing products is as much as the new products, right? So, getting this real-time, Input is hugely beneficial.um, Do you have anything else on the topic? Otherwise, I was going to switch it.

Anne - Be only the only thing I think that you're hitting on probably transitioning into our next topic a little bit but that data piece is how you then convince stakeholders right.That this is an important initiative, that requires investment and that will generate the kind of return on investment that every organization is looking for. And so, that's huge. That's a huge piece of the change management part as well.

Kay - Yeah. Can you speak about change management before and after AI because even after AI, it still affects the business processes before doing AI? It's a lot of convincing and looking at the data and being able to substantiate it. Yeah. Can you speak a little bit more about it?

Anne - Absolutely. Yeah. So I think you know if we look at the example that I gave earlier you know it's It started with a team, you know, the team of experts, whether it's your Regional experts or your service engineering kind of expertise, experts on the hunches that they have, right? So they have these questions. Is this the p.m. cycle that's been redefined? Is it the one that is optimal for the device? Is it doing what's intended? Right? Is it reducing the number of Service events later and I think just starting with that, question knows, And was that had been in place for a long time? Hadn't gotten the right level of attention or investment. Because again R&d didn't have any data to say, why would we believe this? You know, this is what was predefined? Why change, what's working? You know we might introduce more issues, we don't want want to do that and those are all valid concerns, right? And so what is needed then is to get that data, right?

And we ended up in a fortuitous situation where we Had a compare group and a control group that we could look at and basically, and I think any company could do that if they could create their own, their compare group, right? If you could take their requirement away in one area and then wait a year and then see what the data showed. We happen to have historical data, which was helpful. And so we were able to pretty quickly in two to three months. Compared those groups and take a look at what is the data show. And with the Data. Then you have evidence to go back to the right, stakeholder groups within R&D and it's going to be leadership, right? Because that's where the investment is required, both in terms of time from their teams and then also the return on investment. And so you want to show compare those two things and say this is, does this make sense to go after most of the time within service it will because even if you don't generate the returns in one year if you look at the lifetime of that equipment, You're going to generate it into, or multiple five years, right? And it will pay off and so. So that's kind of the direction that we looked and luckily in this case it was very clear well if we remove this p.m. light and here's the implication of that I think with some other types of interventions in might be a little more complex to say, well what do we do about this problem? If it's one particular part that seems to be failing more than another does that require Design, or does that require a replacement strategy? I think that could be a little more complex with some of the other problems. But the returns could be even greater right for something like that. And so, that's kind of how we
moved that particular project forward. It had a pretty clear outcome, the data was convincing. And so the question just became like when can we do this, right? Not should we be doing this?

Kay - Yeah, that's perfect. So essentially what you have given for any leaders, service leaders is a framework to do our think about AI Projects based on our joint experience together. So if I me summarize the kind of steps that you have been guiding us through, you started with a hunch you looked at what is the data that we have to substantiate that. Conch and where are the most efficiencies that can be improved and what information do I need? So it's also coming up with a clear deliverable at the end that needs to be convinced for the change management portion. There was a so that determined a clear outcome for the project and then ongoing, how do you continue that change management and the steps that need to happen? To create, you know, come up with that review and do this periodically. So did I summarize this correctly?

Anne - Yes and I think, you know what you would like to do with the proof of concept. Like, this is convincing the right people within your organization of the power of this kind of approach, right? So that then it becomes something that the next time it's more Blessed. Right? There's less convincing that needs to be done around the model and the data and things like that. If you can get some buying early on and show the proof that that project worked, then that creates kind of a framework and a roadmap to continue this type of improvement down the road. Right? And so that was the vision for us let's pick something very tangible that we can use to develop a model. Internally how to use AI insights to improve service the fish.

Kay - That's awesome. That's awesome. Thank you so much for your time. I think this is very very helpful from a service standpoint, for leaders to be able to start implementing AI within their teams and create that feedback loop and that voice of the customer to the R&D teams. Thank you for your time and for continuing. The Discussion and look forward to sleeping more benefits.

Anne - Thank you for the opportunity.