Good afternoon, everyone, I think. Alright. So, first of all, um thank you for joining us. Um thank you for having us us back, some of us back uh in our second year of uh re invent um and happy thanksgiving to everybody. Thank you for, for being here the week after thanksgiving.
Um this is gonna be, this is gonna be a fun event, I promise. Um and we're gonna talk about some real meaningful things uh that health healthcare organizations are doing uh leveraging technology as an enabler as a business transformation enabler. So we're, we're in for a good discussion.
We'll, we'll try and we'll try and keep about 10 minutes for questions. Um and I think this is, i, i want to make this interactive and keep and have some q and a at the end too. Uh so feel free to uh I think maybe if there's a mic come, come up to the mic and, and, and ask the questions.
So why don't we, why don't we begin with some introductions? Um jeff, i'm gonna ask you first since you're at the end. Uh why don't you um just speak a little bit about yourself, your organization and, um, how you, how you got here in the second year in a row? Clearly, i didn't burn something down the first year.
Um, hi, I'm Jeff Hook. I'm the chief information officer for a uh uh community healthcare organization called Nance Health. We uh practice in both Connecticut and New York in the Hudson Valley in the western side of uh Connecticut. As a chief information officer. I've been with the organization about five years and I got here by seeing the need to stop transporting data, but to transport logic to our data. And so it was very important for me and for my organization to start to take data as a very valuable asset, quantify it and put it in a single place.
Wonderful Elizabeth.
Um hi, everybody. I'm Elizabeth Estes. I'm the executive director of the Open Source Imaging Consortium, also known as OIC. Um and we work right now primarily in rare lung diseases. So how many of you have ever heard of Idiopathic Pulmonary Fibrosis? Couple hands like two or three hands, maybe four hands in the room kills as many people in the United States as breast cancer and no one's ever heard of it.
Um so what we believe is that technology can be used in an open source fashion. We can, we're the largest global database for uh Idiopathic Pulmonary Fibrosis in the world. Um and we have data from all over the world and what we believe is that we can use technology and use machine learning to do diagnose sooner because right now it's a 26 month path to diagnosis. Imagine all of a sudden you get a cough and you go and 26 months later, you finally get told you have a fatal disease and your first question is going to be, how long do I have? And the doctor's answer is we have no idea.
We believe we can use technology to help figure out progression who will progress, who will be a slow progressor and a rapid progressor. And we can do better than that. And then also response to therapy right now, there's only two drugs. Um and that's sinful in the world that we live in. And so the goal of OIC is to use data and great minds. It's sponsored by all sorts of pharma companies, imaging companies and all sorts of really cool start up and medium sized AI companies around the world. And we believe we can make a radical difference on behalf of these patients and their families and their caregivers.
Um I'm Yigal Oren. I'm the senior director of software engineering for Memorial Sloan Catering Cancer Center in the New York uh area. Um this is the first cancer center in the world. It was founded in 1884 when cancer was a very strange kind of word in the first hospital by the created had a round walls because they thought that the cancer virus has to stay in corners. So if they have round walls, it will eliminate contamination going to today a bit kind of different.
I started working in Memorial Sloan about 3.5 years ago. Um starting to bring technology that I work a lot with in my previous life, working for financial industry and trading systems to bring more real time, more faster to market type of attitude to the work that is being done in health care.
Um a big part of it is to try to bring together the different arm of the institution. We have the hospital that deals with cancer. We have a very big research arm that deals with clinical trials, research on new drugs, research on a new type of diseases, mutation of genes and so on. And we have also an educational part and bringing those three together really is what made Memorial Sloan to be the number one cancer center in the world with regards to the capabilities, the innovation and the new things that come from there.
A big part of it is how do we change the whole very siloed infrastructure and the way we do work, bring it together, bring automation, change the culture, not just the technology of the organization to bring them to the cloud, to bring them to AI. And to take advantage that Elizabeth was saying of new technologies to better identify diseases, better treat them. And then better prolong the life of the patients that we work with.
Just, just a quick background on me on myself. Uh Abbas Zaidi, I'm um i, i live in Boston. Uh I served uh healthcare and life sciences clients for a number of years. Uh growing up at Mass General Brigham and another cancer center, Dana Farber and uh and, and Optum for before, before coming to PwC about three years ago.
So, um again, thank you for joining us. Let me, let me kick this off with uh Elizabeth. I'm gonna, i'm gonna ask you to kick us off. Uh you are your organization OIC tremendous organization um nonprofit um focuses on rare diseases, specifically um idiopathic lung disease, um interstitial lung disease. How, what, what, how did this, how did this even happen? How did this come to be and how have you used sort of technology as an enabler to drive uh what your, your vision and mission for the organization?
I have a really heartwarming story about the very beginnings of OIC, a pulmonologist, radiologist, and industry executive walk into a bar true story after an American Thoracic Society meeting. And they said, you know, this was five years ago, almost six. Now, we believe we can use for technology to level the playing field in this disease because with Idiopathic Pulmonary Fibrosis, right, it's a, it's a variable diagnosis, there's variability in the radiology reads and unless you're in an academic medical center, chances are no offense to the radiologist in the hinterland, like where I grew up in the middle of Illinois, but they have never seen this disease because it's rare. So they don't know what they're looking at.
And so a patient would be sent home. And so this group got together and said we should try to do this, we should build a global consortium and we should give away open source what we learned to the world. And so they gave me a call and said, would you consider running this? And I said, I'll do it on an interim basis. You can say how good I am at keeping my word, but we're getting there in five years we've built, so we've built a great database and how we've used technology is we started PwC, help us a great deal um in the beginning and we started and we built it and now we've just switched over to the cloud with our partner Vita and PwC helping along the way and being a part of all of this.
Um the hardest part for us was the data that we're getting is from all over the world. So not just in the US. So imagine they call the disease something different, right? We know now there's nothing that we have at the moment that can automatically change those disease uh names and how they call them into something into curated data. It was really important that it not just be imaging data that we make it enriched data for our artificial intelligence and machine learning teams all over the world who are using it by adding the clinical aspects to it.
And you always say to me this is a small but mighty database and it is so to give it perspective, the largest database for Idiopathic Pulmonary Fibrosis in the world is around 3000. Right now, we have 25,000. So for those of you data people in the room, you're going 25,000, it might as well be 25 million because this disease is rare. Now, the question is, is it rare or have we just not diagnosed it enough? Right. And we think there's a little bit of both.
So we're using the technology now to bring the data in to anonymize it, using AWS the cloud to bring it in to anonymize it, drag and drop. It's amazing with a case. We're annotating it so that it becomes more enriched and more enriched as we go on. And really, we started in the middle of our challenges. Our first was earlier diagnosis that's low hanging fruit because IPF looks very distinctive on a pattern. It looks like honeycombing looks like honeycombs in the bottom of the lung. So that's kind of an easy we started with progression. Is this patient going to live or die? And can we figure out which patient is going to live and die, live or die? And how long are they going to have? And so we've made a great deal of progress with algorithms for mortality to do that using machine learning tools.
Um and then as we move forward, we're now looking at adding other rare lung diseases, sarcoidosis alpha one, an trypsin. And importantly, and most importantly, we're starting to bring in lung cancer screens because smoking is, is actually a risk factor for this. So the technology for us is input, right? And a lot has to happen with that input crfs creating clinical guidelines, understanding what's important for the machine learners, be willing to move and pivot quickly if you need to do that.
And then on the output side, we have algorithms that our algorithm companies have already created. We don't know who the lead horse is yet. We, we do challenges. We've done a Cagle challenge, we're gonna do some other challenges um to try to figure that out. But the the technology in this space and I say this to all of you, we believe we can prove a model in rare disease that can be applied to any rare disease. They deserve the same benefits that someone who's suffering from lung cancer or breast cancer or prostate cancer deserve. And we can get there, the tools are there. We just have to have the attention and the focus and then be able to bring in data and really curate it in the right way.
So for us it's a game changer for these patients. It's a game changer. Um, and we've learned a lot along the way and made a lot of mistakes over the last five years, which I'm totally happy to share with any of you who want to know what we did wrong. We'll be more than happy to tell you that.
No, that's great. Elizabeth and, and, and for all the health care and data geeks out there, I know, I know there's some out there. I see Beth Folder in the in the audience right there. What's amazing about OIC is that they've integrated imaging data and clinical data, truly integrated data set that's actually not common in the industry. I i find it very, i found it very rare to have a truly integrated data set that tells you more and we'll come to what's next. But that, that's, that's the impressive part. So that's amazing.
Yal, let me turn to you. So Sloan Kettering, you know, amazing institution, right? Maybe not as good as Dana for no, but, but how, how have you, how have you thought about pathology applications, protocol, information management systems, um cohort groups for academic researchers for clinicians, administrators and employees, right? Uh i think jeff, you mentioned digital, digital clinicians, digital workforce, digital, digital um experiences for patients, right? How have you brought, how have you thought about applications in your, in your area?
So one of the things that really starts us in the right way is to get an understanding to the technology because we are here, we're talking technology, technology is not the main business of where we are. It's really the hospital and the and the patient and the research has done around it. Technology is supposed to serve those different groups to get them to be better, to be able to provide better care, better services and to find better cures.
So, what we're trying to do is we're trying to concentrate and see how we can use the current technology and the current product and the current capabilities to do that cloud is one aspect of that is something that help us and move the whole process to be closer to the different function that we want to serve and also to provide better time to market, to be able to give the different department, the capabilities to iterate fast, to find the different solutions to move forward.
And what we found working with PwC for the last year and a half that to do that requires a lot of grunt work. So we had all the things around it that we are talking about dev ops that needs to be in place, the data structure and the databases that needs to be in place, the data in the whole interaction between the data and the systems needs to be in place. And we worked for for almost a year and a half with the team that some of them are here and we have a great relationship with the MSK team and we put all of that together.
So we got to a point that the real function can now iterate in a much faster rate. So we can get different deployment of changes to systems of capabilities out every two weeks, every four weeks, we can get them out there that before it used to take months. Um and that threw a lot of automation and a lot of capabilities that we brought in together to make that happen
When we talk about the different capabilities around, how do we deal with uh AI and uh the cloud, I'll give one example for there's a very serious process that our research department is running and we internally have what we call scientific data center. So we have GPUs, but we have a limited amount of GPUs. So when they run their model, it takes them about 80 days to run through the whole model to get the results that they want. And to get to a point that they start iterating on what they got out of it by exploding into AWS. We can take this whole process, use 100 or more GPUs at once, run it. And we found out that we can run it instead of 80 or 90 days, we can run it in a week or two. So just think in your head of what does it mean? To a researcher that they can get the results of their model in a week instead of three months. So we can iterate now, eight times a year, instead of maybe four or three times a year and find different solutions and different capabilities to bring that to play other things that what does it mean to the patient? Does it mean the patients are getting the solutions quicker? Correct.
So, the whole thing about absolutely, the drug design, the capability to find the right drugs, the capability to find the mutations on the genes. And when you look at databases and you look at the gene, it's huge to try to find for each patient and tailor the different treatment to that patient. Whether it's the age, the the origin, the race, uh the different rotation that, that that patient has all the different side effects that they have on that drug and bring it all together. That's a huge effort requires a lot of horsepower, a lot of memory, a lot of collection of data. And that is what makes this whole thing a lot easier when you run into the capabilities that the cloud allows you to do because we can expand everything, get all the stuff run for two weeks, i can shut it down, go do the calculation, do what i need to do and then come back.
So this whole iteration around helping research is, is 11 thing the other thing that uh Elizabeth was mentioning is looking at imaging and looking at how image and processing of an image using AI can help diagnose better and then treat better. So what we saw through a lot of research that we're running, if you take a pathology slide or radio image, and you just take the doctor or the physician to look through it to identify what they want. You get a certain percentage of accuracy. If you run the AI image on that, you get similar accuracy. But when you run it combined, you get between 10 and 20% improve on both identifying the tumor and second finding the right way to handle it.
So if you just take that and you think what does it mean to a person that has cancer to know that they can identify it earlier, know exactly what it is and then also tailor it to the right cure to bring that home and get that person and the family of that person to be treated in a better way. Speed to value means very different things in health care than it does other industries. It means life or death, right? For, I mean, you know, heaven forbid anyone in this room has, has, has gone through cancer or a family member. It's, it's literally speed to value and and what I heard Yigal was cloud has given you the opportunity for not just performance security. I think you've used that term before. Right. But, but and scalability, right? So researchers can have access to data, faster, access to insights faster, right? Not just data, but the insights so that, you know, yes, you know, clinical trials can happen faster. And yes, it takes 2.2 and six cents billion on average to make a drug. Right. So, so certainly I think some of you you've already seen some of those benefits to come to fruition.
Yeah, absolutely. And, and one of the things just maybe to add to that is I worked in financial industry as I mentioned before. So there, the time to market is how fast can I do a trade? So it's just, oh, somebody made a little bit more money or less money or was able to do the trade. People's life and their family and you can see that and what we see with a lot of people that I work with in MSK, many of them as family members that went through that in MSK, including me myself because my wife had to be treated there and you know what it means, you know, what it means to wait another day for a result. You know what it means to, to that the doctor needs more time because they don't have the right tools to analyze things. So it makes a huge difference both on the person and on the family.
I was just going to chime in from a, uh wonderful research organizations represented um as an academic medical center, uh the speed to value is maybe the difference between walking out of the hospital or being wheeled out of the hospital. And so minutes matter stroke. Um and so resilience of cloud resilience of data, the analytics to help the uh rate biologists and the treating physicians. It's not just speed to value its speed to life, speed to functionality. And if we fail either in our identification or we fail in having redundancy or resilience, then that could be your loved one, went to the same hospital. But as a matter of moments, that could be the difference between that blood clot being removed in time or not. And so those are some of the things that we think about um with our technology, with our technology decisions, with our implementations, with our analytics. And so that's, that's part of the thrill as well as some of the burden that we uh uh commiserate over. And so that speed to value is very, very, very important as a lot of you will experience in your lives.
So jeff come last but not least. I mean, so new vance health, right? As you said, connecticut and new york seven hospital regional system. But wow, I mean cloud, you started this journey way before you know, some health systems even got into this. What what how have you used the cloud to just AWS to just drive change at your organization I love to hear more.
Sure. I did anyone watch this morning or participate in this morning's keynote um start with security, start with security, start with security, start with security, start with security. Um it's location, location, location. Oh, that's real estate. Um so we started with a program that said we need secure. Um a lot of skeptical concerns, community members. I mean, we were talking about um is anyone familiar with kind of the health care industry? PHI I, I raise a hand just kinda um so you all know that IP address is now PHI um yeah, that's pretty awesome, right? Um so, uh again, so we started with security regulatory risk and said, what type of system do we need to build? Um and if you've been in the industry for a while, we usually deal with 5 to 10 to 20 year old technology, we bring data that's inconsistent, we slop it from one place to the next. And so the vision was we need a secure place to start to bring all of our data, break down the silos. And then we're going to bring logic and systems to the data rather than moving the data around to those systems. And so that was a journey that we took to both set up a secure foundation, secure organization, dip our toes into the cloud. What i didn't mention or boss didn't mention is we're a five year old organization. Um and it grew out of, you know, some community hospitals to a small system, to a merger that created a larger system. Again, we're still a modest size, but we didn't necessarily have the pedigree and the experience in our teams to start to do those things. And so that's where we said we need a secure platform. We need help building a cloud platform. We need to have a very robust analytics program and then we need to set a visualization and integrate that governance into our organization.
Um and through the support of our board through our ceo uh we actually have a fully integrated analytics program that drives our strategic goals, drives our day to day goals and creates transparency from board member view down to unit, you know, nursing, unit manager, view of your key responsibilities or your role in certain um goals. And so we're very excited about that. We're not quite as far along as some of the AI but that is also uh I've called this the trojan horse strategy. It was let's get our data in a single place. Then we can start to work on analytics. We can then work on app development and we can start to work on AI. And that's really the next chapter that we're taking. Again, very excited. But again, we for myself, i fight a lot of fires every day. It could be whether it's getting the imaging team ready, it could be uh a physician really needs a new computer. Why did he get denied? You know, it could be all kinds of fun things.
Um and so it's exciting to have this platform, these tools and then to partner and learn from peers. Um sitting at this on this platform with me and uh talking about that. So again, i love what i do. I love what we bring. That's great. Thank you.
So, jeff, i'm gonna, i'm gonna dive a little deeper for the folks in the audience. It seems to be a large health care contingent here, right? So i'm gonna, i'm gonna dive a little deeper into some of the analytics that what you've, you've truly liberated the data uh and, and driving an analytics mindset in the organization, right? So, um and, and so some of the analytics that you've driven and uh and that beth can keep me honest here. Uh but you know, everything from patient experience and digital front door, uh patient safety and quality uh financial metrics um value based care. Where do you, where would you suggest someone in a, in a, in a similar situation? Where would they start? Where would they get the most? Not just interesting but, but speed to value and value for money in terms of an investment there for analytics.
First call the boss. Um that he's gonna pay me 20 bucks for that later. Um no, i think speeds of value. The most important thing is again, did i mention security? Get that secure system? Understand what it is? Get the concepts there. Also, your team is critical and i know i'm probably speaking with you guys know this stuff cold from AWS. We didn't, we did a whole, is it AWS? Is it, is it google? Where do we go? And so it is understand what platforms you need, understand what you want to achieve, make sure you have a secure environment. And then from there, it's pick what aligns with the needs of your organization. You can talk about the quadruple aim. You can talk about patient satisfaction, you can talk about, i wouldn't start with financials. That's pretty standard and fair. But it's about what hasn't been exposed, but what is not that hard to get there. And we looked at patient sat as our very first test of change and through a satisfaction element, it came from a single vendor. It came from this whole organization and it just was a great proof of concept to bring that data, get it exposed to our ceo, get it exposed to our chief quality officer and every president of our hospitals and our medical group, et cetera. And then we could cascade it down very quickly. And it took us a while to get data governance there as well. Everybody thought they knew data governance and everybody thought they were an expert on what data we should have. And so as we went through this and realized data is very expensive and very valuable. And so going through those steps to go through data governance and start to use the term ROI and which priorities you have and then you can start to ask your own organization and then it creates that really sea change, that the right thing is getting prioritized first. And ideally, you build a huge case for trust between yourself and your ceo your board and the rest of the executive team, elizabeth, i think you and carmela have a lot of experience in that.
Do you want to share your, you know, it's interesting. He talks about the analytics that he gets at a hospital system. So my job is to travel around and get people like this to help and give me data so that we can make our database broader, right? As you know, it's so funny. I said to somebody man, if i had a nickel for every time i heard the word generative today, i'd have as much money as bezos, right? Notice, we haven't said that word. Yes, we haven't yet. But i'm going to say this in order for those to work, you need a complete diverse data set, right? It cannot be homogeneous data set because you're going to implicitly add bias to those to those generative. There goes, i'm going to give a nickel back a i that you're using. But what's what's really important about what he said. So and remember i come at this from a different vantage point. I'm begging them for data to help us make our database broader. But some of the analytics that he's getting the bioinformatics that we can get from hospital systems. Bioinformatics is the most underutilized i think tool in health care research today
And I'll give you an example. The disease state that I work on mostly - pulmonary fibrosis - it's a diagnosis of elimination; takes 26 months for them to get diagnosed. So they go to the doctor and the doctor, they go to a hospital system like New Vance in Connecticut. And they're going to get diagnosed and they misdiagnosed, they're going to get given antibiotics and they're going to get steroids. And then 26 months later, they're going to get, you know, thrown to a pulmonologist who's going to now tell them that they have this fatal disease.
If I can find patterns in those bioinformatics on the ICD nine and 10 codes, I can go out and do an educational campaign for every doctor in the United States and every doctor in the world. If you have a patient that shows up and this is what happens to them, get them referred sooner, right? So you're grabbing analytics that from a researcher is golden. And you know, I was taught this by my friend Atul Butte from U CS F in San Francisco. He's a brilliant bio informatic guy who went through all of the UC system and found out that they were treating diabetes like 48 different ways, right? And they went through and she said, why, why it was all because that's how the clinicians wanted to treat it. And so they tried to get, tried to get some consistency.
So those analytics that you're getting, that you say, ok, we're going to get patient satisfaction faction. But you're also getting these other ones from a research perspective. That's golden, that's absolutely golden. And that's things that, that we're interested in on our side.
Yeah, getting data has been um do I dare tell the story a challenge? I, when we started this five years ago, I always tell people, especially in other countries that I was the either the naive or arrogant American and you all pick. I thought 15,000 scans, which was our original goal. I was like, how hard could that be? Like 15,000 scans I can get this in a year.
Um so I started in the US and I went to hospital systems and the very first one in the Midwest said to me, Elizabeth, we'll give you 300 IPF scans and clinical data, $50 a scan. And I was like, wait, I have to pay for this like literally and I'm like, oh ok, 15,000, I think we can do that. We're a nonprofit, right? I asked for the very same thing in the southern part of the US. The same exact data. And I was told $250 to scan. I went to the east, the northeast, same exact data. And I was told 700 to scan. And then I went to a venerable institution which I shall not name. And for 500 scans, they wanted a million dollars. And I said, no, I'm out.
So I flew to Europe and I got to Europe and I, I remember vividly, somebody said, Elizabeth, we'll give you the scans. I'm just wondering, are you good with GDPR? I'm like, oh boy, eight months, it took us eight months. We did an entire uh project with a, a law firm called Wilmer Hale in Frankfurt who looked at exactly what we were gonna do and how we were going to get the data and eight months later back to that institution in, in Europe and they said we'll give it to you. No problem. And then another one gave it to me and then another one gave it to us and we just had to guarantee security, security, security, right?
We had to make sure that there was no reasonable because GDPR, if you're all getting data, go with the GDPR because hi a, you're good to go there. GDPR is the gold standard, right? And, and there's no reasonable way that it can be reid and we absolutely do that. But I will say this when you're dealing with a rare disease, patient privacy is for the, well, if you're gonna die, you'll give your data.
So we need to come up with some really interesting, secure cool ways for these patients to be involved in their own progress of their own diseases. And so that's OSI's next step in what we're doing, getting the data and gathering the data and then curating the data. You guys Vita, I mean, how we've curated this data. I will tell you that we did a very, you guys are all going to laugh very common thing. When we set up the database, we were letting perfect be the enemy of good. And so we went back and said, nope, nope, nope, we want to do minimum viable product. So what the radiologist and pulmonologists thought we needed and our machine learning lead from University College London thought we needed sure enough, we went with really little scale and then we started adding more and more and more. We built a plan and a CRF, a case report form of all the clinical data we needed. And then we expanded that as we went on and that's what's made this data so valuable and but it's, it was a learning experience for sure.
So two things I heard it's hard, it's hard, right? The data pipelines getting all the data, it's hard, right? So we're not going to talk about the G word. No, no, no, no, no nickel here, but it's, it's hard getting. So speaking of hard Yigal, I want to talk about MSK Mind because I think if there's a game changer and it's hard love for you to know. I mean, this, this whole conversation is about operational and administrative uh not just migration but redevelopment. Refactoring relat into the cloud. Like what, what are your thoughts on? What have you done so far and how have you done it and would love to talk about the future as well. But let's start with MS G.
So, so one of the thing talking about the difficulties, as you said, there's multiple layers of difficulties, there's difficulties in the technology side because we're talking about serious big amount of data. When we talk about a cancer institute like MSK, when you take all the different images and you need to scan them. Each image is huge because you need it to be with hives of a resolution. You need to make sure that the pathologist will agree to yank him out of or her out of the of the microscope and be willing to look at that in the screen. You want to be able to bring it to the screen in a speed that is good enough for them to accept it. All of that. You get, you can find the right technology solution, right? You get high speed disk, you get high speed network and but that's all local. The minute you want to share there is a problem. How do you share such a huge amount of data. Where do you do it? And that's where things like cloud come into play because MSK is again and similar to Elizabeth is a non for profit.
One of the things that I like about the place and I'm so proud to be part of it is that we focus about the patient. Our mission is to detect and cure cancer and to do it in the best way and the, the fastest way possible. And that's where all the physician, the clinician and the researchers are really putting the mind into and they, they don't mind about having their name out there. Of course, they will have it and they don't care to keep secrets. So it will be only us what I love about this institution. They want to share, they want to be able to share with the community with the, it's exceedingly, it's like, for example, one of the things that MSK the research department work on is to find the different cancers with the gene mutation and, and the, the cure impact and the side effects and they created a huge database that there's no parallel to it. It's called OncoKB. They put it on the web. Any of you can go now and do OncoKB and, and look at that and you can find the, the different disease, you can find a different uh mutation because they want people to be able to take that and maybe help someone and and so, so that kind of capability requires a lot of technology behind it, it requires the technology to process it. As as you said, the whole MSK Mind is is an initiative that we're doing with research that brings together the different type of research that is being done and trying to combine that type of effort to collaborate on data, to collaborate on compute, to collaborate on models that are being created to analyze the data and be able to share it.
One additional thing that we do, for example, is trying to go before, before cancer starts. So we have a whole process now that is uh identify risk and people. So any, any person can go now to MSK uh dot org. There is an a survey that you can go through and say, you know, i have this kind of a cancer in the family, this is the relationship i smoke. I do and, and they will give you an assessment of what you should do that next. What type of should you go to to do testing? Should you see a physician? Should you do a anything and go and work with that and that's open to anybody.
So, and, and going back to that security issue, we always have to keep that in mind because the last thing you want is to be able to have a patient or a person that engages with your organization name or information. Get out an eagle doesn't look good in orange. Yeah. And, and i, i tried that on it didn't look so, so security is very important. And then we talked about um the data, if i need to bring that data, those are tens and tens of petabytes of data. If i need to put that somewhere, i need the right capability from the technology point of view. I need to access control both from security and availability of the data. I need the capability to share it externally.
Um we, what we are doing, for example, with the hospital, we're doing a lot of second opinion uh because we are not available uh outside of the east coast at, at this moment. Uh we provide those services, international and national so people can come and say i, i was identified as xy and z. I want to get an opinion from MSK because I know the name of MSK and i, i appreciate the capability to um find out from a physician in MSK, what they think. So they need to upload data, they need to upload test, they need to upload imaging, they need to upload uh pathology reports, they need to bring all that stuff out. Again, you need a place where you can do it and enable clients and an easy way to do it. And also to make sure that you have all the security around it that is available only to the people that should see it and so on.
That's when you look at that, it goes across all the different areas. So even though we are a cancer institute, so you say other hospitals deals with a lot of different areas and department, it still has a lot of different, separate and different clinicians and research areas that look at things and they're not always in agreement about how things should be done. So what i find from a technology point of view when we work on those things is we need to create that kind of environment or marketplace for everybody to first trust us that we provide the right services. And second that we are resilient enough that we are available that we will give them what they need when they need to in a way that works both financially and from a business point of view well for them. And that's something that takes a while to create because physicians and the research people like to operate on their own and, and everybody, if you work in an environment like that, it's not easy to her, all those cats because a certain area of of research gets a grant, they can start their own project, they get their own aws account and suddenly they have their own regions here and they run on there and, and then you have to go and chase it and find out what happened. Because after a few months, the person that had the grad and created the system left and now they're coming back, they say, wait a minute, the system is not working.
So it sounds like we need some cloud governance. So. Right. Right. So, so, so from a i remember and abbas thank you for bringing me for a repeat appearance this year, i remember last year i spoke with some folks afterwards and there's some entrepreneurs, some developers, all of you, or at least i saw a lot of hands about health care. And i think one of the things and just as i hear my research colleagues talk about research, there's a lot of brain power going in. There's a lot of data being associated there. But i think there's also as you take in context as conference, the tool sets.
Um what i would also implore is not just take the analytics but use the analytics to set up the next best action. And that really needs to be some of the mindset that we take either with the tools with the analytics because analytics are no good if you solve cancer but don't do anything with it or we've researched the hell out of this thing, but i can't get anybody to use it. And so whether it's for patient sat correlation and again, i use patient satisfaction. We've done, we have six pillars that we set our strategy around. We have score cards. And i think we somewhere around 300 different unique analytics around those six pillars all built out in the last two years. And so what is an x expects action now that you've found this data? You have the insight, what are you helping people do? And i think that's where if i were to look to the future or look to where we're taking this. Now, i've built the safe environment. We brought the data together, we have a picture and people say, ah we have a problem and then it becomes now what? And so it's using the tool set that we're talking about and i will use the man. I'm tired of hearing different people tell me how a i is going to solve my problem
Like, thank you. Thank you for bringing that to me. It's so much easier. Now, I didn't think of that over the past three years, but from those elements, we're prepared for that. If you have your data organized, if you understand, if you had autonomy taxonomy, excuse me. Uh you know, the challenges you're going to run into are the how many different ways did we document c section in our organization? It was c dash section c period, section c space c underscore. And so those are some of the challenges that we run into with taking the data, we get it cleaned up. But then what's the next best action? I think that's where we look at. Whether it's q again, real excited about that. Getting q connected to connect. Really excited about that. Again, that's just me, ner, what about kiper satellites? You know, there are, there's probably some consumers of mine that would love to have satellites because they don't want cell towers and they don't want cable, but they live in podunk. So maybe we'll get kiper in there to sign me up. Uh but no, it's what is the next best action? How do we solve problems for people? And that's the challenge that we have to look at um across the board. I i appreciate the tools and the power that we have. Yeah. And I think, you know, for us, it's the same thing, right? For us, we say that's great. We can build an algorithm that says which patients going to progress or which patient is going to die. But the question is, you know, how do we get that to, to the clinician? And how do we make it matter clinically? That's our job, right? So what is the technology and how does the cloud help us do that? We've got to start here so that we can give those physicians? You know, and, and honestly, the pharma companies, i get people, you know, a lot of naysayers or cynics will say to me, elizabeth, you're just trying to get pharma to create new drugs. And I say, yeah, there's two, my sister died of a ls. There were two drugs. How is that possible in a disease that we've known about since joe dimaggio and we all dump water over our heads for the ice bucket challenge. The ecosystem isn't set up for speed. Right. The technology is there. It's like we've got, it's like we've got cars that go 100 miles an hour, but we're still driving on gravel roads. It's our job to figure out how to make those ecosystems right, and get there quicker. And it's a, it's a big, it's a big passion obviously for me after doing this for five years because seeing it from the inside out and i wasn't in the health care space five years ago. So I've learned a heck of a lot.
One of the things that really, we're talking about the outcome of technology that we implement analytics, we implement a i we get to. But the, the one thing that a lot of time we forget is really what do you need to make that work to make that really happen? And that's part of the thing that is the kind of the dark side that enables us to do that because what we found out that to took one application and migrate it from where it is and take it to the cloud in the way we want to take the first one, took us about nine months because we had to find out, oh, this is not there or the terraform was not created in the right way. We didn't have the right thing or j frog was not all the things that nobody wants to hear about. If we don't create that in a way that is like a pipeline that you can say, you know, i don't want to think, i don't need to think about it. I want to use a i, how do i use it? And then you can take it and use it and really take the benefit from that fast is key. What we found in in about a year and a half that we went from a process of nine months to take an application and take the development team and kind of stepped down. And i see the smile on the team that that really worked with me to do that and to come to a place that we can do it in two weeks. And suddenly the application is, i can build an environment, just click on the button in 10 minutes. My development environment is up and my database is permission and i can start really putting data and run what i want to run there. That's a key thing that sometimes people skip and sometimes people don't have the patience to see that through. And what we found is it comes to a lot of other things at the end. One of them is the speed. Of course, the second thing is resiliency is the capability that if your environment goes down, but click on a button in five minutes, the environment is back up or that your data is back up if you want to save money because your organization is not for profit and we don't have lots of money. You can drop your test and your t environment during the weekend and not use them because you know that monday morning you click on it and before you come with your coffee, the environment is back up and you run it, all those kind of things are not there. If what i call the little people running around, the, the the screen really put all those things together and really combine them and work them and it's hard work. It's a lot of work that requires a lot of attention to details and then you have to change the culture of your team. But I would also say, and, and you know, again, i look at this, i'm going to take the helicopter up a minute because he's talking about that from, from the perspective of, you know, ms k, right? And you're talking about it from the perspective of nuance, i say we even need to get standards globally that are similar, right? I have no idea how to do that. But i, we talked about this with the rsna. I want the rsnas going on right now this week in chicago, i wish there were standards of how they take a standard chest ct or hr ct for this disease because they're all over the place or i wish there were standards on, we talk about the taxonomy, right? We, we've got to figure out a way to come together and collaborate and you know, sic is partly that, but it would be great if hospital systems from the us and hospital systems, like we have data from asia and south america and, and you know, all over the world and if we could come together and say, how are you doing the hr cts, let's be consistent because i think we've got these tools now that allow us to create algorithms that work on real world data around the world, not just first world data. And that's super important, but we've got to, you know, it's just theory, we change the output, change the input. We've got to get more consistency. Like, are you guys doing the same thing on taxonomy? Are you doing? Oh, definitely, i promise you, i promise you, you would know that my point is is that i get your pain point and i get your pain point and then from solving disease and using a i and machine learning to really make progress. We've got to have a higher level set of standards and a higher level set of consistency. There's my soap box moment and, and i'll reply with health care is local and you know, that's, that really creates a very interesting contrast of how we're trying to solve things and how difficult it is because i have to meet the needs of my local organization and i very much care about the well being of the global health, but i have to make sure that i'm meeting the needs of new band's health constituents and those who we consider our patients and consumers. And so that's, you know, setting our priorities or even set differently just based on how we operate. And I think that's a challenge that maybe policy can help. But i'll leave policy and some of those elements out of this conversation. All right, i'll put, i'll put you out of policy. Thank you. I appreciate all that. Otherwise i'll get, i'll get in trouble here. But i'm hearing the theme here before i get into what's next. I see some word convergence. I do see convergence. You know, elizabeth, you deal with not just data but pharma researchers, right? Jeff deals with community health based system, um lots of pc ps transactions. You got oncology, second opinions right there, there's, there has to be some convergence for, you know, some of it to work easier for the patient, right. Eric topple wrote that book 10 years ago. I think, i don't know what year but you probably know it. Creative destruction of medicine. And that was the number one thing he said was going to come change. Medicine was the convergence. He's right. It's just taken us a little longer and we've got a long way to go before we finally get there, but i absolutely agree with.
Right. So this morning for those who saw the keynote adamski talked about, you know, what, what he's looking at the future, what he's excited about, right? Um he's excited about and he showed that picture about sort of foundational models with training and inference and then, and then the middle layer of, of, you know, ms and then an application layer on top of that. And of course, he mentioned project hyper as well, jeff. But um you know, if i were to, i got, let me start with you. I mean the next you know, 18 months, what, what are you most excited about? And how, how are we going to use, you know, the cloud in aws to really drive change at ms k?
Um i think one of the main thing for me is is what i just mentioned about research because i find that uh the cloud has the most benefit for the research side um in multiple areas. One is the speed to process. The second is the tools to, to use the a i and machine learning and derive the data and the analytics they want from that. And one thing that is even sometimes can be more important is the capability to share data, the capability to have additional data come into those models. Because when you analyze the data only for what you see in your hospital, it's one thing. But if i take from india and the hospital from africa. And i bring the data together everything changes, the outcome change, the way you want to treat change. And i think that what is so important and so working and that's where i see the next step that of course, will impact the way we treat and the way we find different cures and medicine.
I'm hearing business model reinvention. Absolutely. E for us, you know, again, myopically for the open source imaging consortium, it's the ease of being able to get data in to the point of i have data coming from all over the world. And you know, with the platform now they drag and drop it anonymize, right? Which is awesome. We have case viewers people, our partners annotate, which is a very important piece for us. Um there's all sorts of the ease of being able to use it for research and then the ease of being able to have the output so that the machine learning community can take it. They can create cohorts based on kernel slice that, you know, kernel size, which who knew they needed that slice thickness, clinical clinical fields that we had no idea they would be as important as they are. So i think for us it's the cloud makes our global database possible. I mean, there's no question about it. And we were on another platform earlier as you know, and it wasn't nearly as user friendly and it wasn't nearly as fast. Um and, and as secure and so from, from our perspective, the game changer really is the fact that we can now be anywhere around the world and they can drag and drop their files, their imaging and everything right into it. And good to go anonym organization, you know, we're good to go. So speed, speed, speed and quality, i would say quantity.
Ok. Jeff, what was the word business model reinvention? I really think that's where we're at. We have a head of steam on analytics. And so it's now, what do we do about it? How do we use it? How do we grab more and assist our organization? I don't know if um you all pay attention, but the provider, health care industry really has gone through some significant expense challenges, particularly around labor. And so that means one, we have less nurses, less physicians and we're going to continue to have less. So we have to be very efficient. We have to be very thoughtful, we have to make their job pleasant rather than miserable. And so it's really taking this data and helping them either make decisions or take some of the burden that they have today in removing that. I know that's a slight deviation from what we've been talking about, but that's what the data is going to tell us. It's going to show us how do we use it? How do we re invent and how do we simplify and so that's next best option or next best action make it. So i don't have to stay up all night documenting, make it. So i don't have to stay up all night thinking about what's best for my patients or responding to text messages that are coming in through my inbox. So that's really what we're focusing on over the next 18 months. It's also still 20% of our gdp amen.
So before we go to questions, one more thing i would add is the cool part about the cloud to me is also, although we haven't set up to take imaging and clinical data now bringing in audio sounds because ipf sounds very differently in a stethoscope than any other lung disease. We're also bringing in unstructured data through n lp clinical notes, right? So the fact that we have these flexibilities that we never had before, i think it's a game changer for research. I really do.