Cerner’s lifesaving sepsis control solution shows the potential of bringing more AI-enabled IoT to the healthcare edge

workingThe next BriefingsDirect intelligent edge adoption benefits discussion focuses on how hospitals are gaining proactive alerts on patients at risk for contracting serious sepsis infections.

An all-too-common affliction for patients around the world, sepsis can be controlled when confronted early using a combination of edge computing and artificial intelligence (AI). Edge sensors, Wi-Fi data networks, and AI solutions help identify at-risk situations so caregivers at hospitals are rapidly alerted to susceptible patients to head-off sepsis episodes and reduce serious illness and death.

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy.

Stay with us now as we hear about this cutting-edge use case that puts AI to good use by outsmarting a deadly infectious scourge with guests Missy Ostendorf, Global Sales and Business Development Practice Manager at Cerner Corp.; Deirdre Stewart, Senior Director and Nursing Executive at Cerner Europe, and Rich Bird, World Wide Industry Marketing Manager for Healthcare and Life sciences at Hewlett Packard Enterprise (HPE). The discussion is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Missy, what are the major trends driving the need to leverage more technology and process improvements in healthcare? When we look at healthcare, what’s driving the need to leverage better technology now?

Missy Ostendorf

Ostendorf

Ostendorf: That’s an easy question to answer. Across all industries resources always drive the need for technology to make things more efficient and cost-conservative — and healthcare is no different.

If we tend to lead more slowly with technology in healthcare, it’s because we don’t have mission-critical risk — we have life-critical risk. And the sepsis algorithm is a great example of that. If a patient turns septic, they have four hours and they can die. So, as you can imagine, that clock ticking is a really big deal in healthcare.

Gardner: And what has changed, Rich, in the nature of the technology that makes it so applicable now to things like this algorithm to intercept sepsis quickly?

Bird: The pace of the change in technology is quite shocking to hospitals. That’s why they can really benefit when two globally recognized organizations such as HPE and Cerner can help them address problems.

cerner logoWhen we look at the demand-spike across the healthcare system, we see that people are living longer with complex long-term conditions. When they come into a hospital, there are points in time when they need the most help.

What [HPE and Cerner] are doing together is understanding how to use this connected technology at the bedside. We can integrate the Internet of Things (IoT) devices that the patients have on them at the bedside, medical devices traditionally not connected automatically but through the humans. The caregivers are now able to use the connected technology to take readings from all of the devices and analyze them at the speed of computers.

So we’re certainly relying on the professionalism, expertise, and the care of the team on the ground, but we’re also helping them with this new level of intelligence. It offers them and the patients more confidence in the fact that their care is being looked at from the people on the ground as well as the technology that’s reading all of their life science indicators flowing into the Cerner applications.

Win against sepsis worldwide 

Gardner: Deirdre, what is new and different about the technology and processes that makes it easier to consume intelligence at the healthcare edge? How are nurses and other caregivers reacting to these new opportunities, such as the algorithm for sepsis?

Deirdre Stewart

Stewart

Stewart: I have seen this growing around the world, having spent a number of years in the Middle East and looking at the sepsis algorithm gain traction in countries like Qatar, UAE, and Saudi Arabia. Now we’re seeing it deployed across Europe, in Ireland, and the UK.

Once nurses and clinicians get over the initial feeling of, “Hang on a second, why is the computer telling me my business? I should know better.” Once they understand how that all happens, they have benefited enormously.

But it’s not just the clinicians who benefit, Dana, it’s the patients. We have documented evidence now. We want to stop patients ever getting to the point of having sepsis. This algorithm and other similar algorithms alert the front-line staff earlier, and that allows us to prevent patients developing sepsis in the first place.

Some of the most impressive figures show the reduction in incidents of sepsis and the increase in the identification of the early sepsis stages, the severe inflammatory response part. When that data is fed back to the doctors and nurses, they understand the importance of such real-time documentation.

I remember in the early days of the electronic medical records; the nurses might be inclined to not do such real-time documentation. But when they understand how the algorithms work within the system to identify anything that is out of place or kilter, it really increases the adoption, and definitely the liking of the system and what it can provide for.

Gardner: Let’s dig into what this system does before we look at some of the implications. Missy, what does the Cerner’s CareAware platform approach do?

Ostendorf: The St. John Sepsis Surveillance Agent looks for early warning signs so that we can save lives. There are three pieces: monitoring, alerting, and then the prescribed intervention.

It goes to what Deirdre was speaking to about the documentation is being done in real-time instead of the previous practice, where a nurse in the intensive care unit (ICU) might have had a piece of paper in her pocket and she would write down, for instance, the patients’ vital signs.

A lot can happen in four hours in the ICU. By having all of the information flow into the electronic medical record we can now have the sepsis agent algorithm continually monitoring that data.

And maybe four hours later she would sit at a computer and put in four hours of vitals from every 15 minutes for that patient. Well, as you can imagine, a lot can happen in four hours in the ICU. By having all of the information flow into the electronic medical record we can now have the sepsis agent algorithm continually monitoring that data.

It surveys the patient’s temperature, heart rate, and glucose level — and if those change and fall outside of safe parameters, it automatically sends alerts to the care team so they can take immediate action. And with that immediate action, they can now change how they are treating that patient. They can give them intravenous antibiotics and fluids, and there is 80 percent to 90 percent improvement in lives saved when you can take that early intervention.

So, we’re changing the game by leveraging the data that was already there, we are just taking advantage of it, and putting it into the hands of the clinicians so that action can be taken early. That’s the most important part. We have been able to actionize the data.

Gardner: Rich, this sounds straightforward, but there is a lot going on to make this happen, to make the edge of where the patient exists able to deliver data, capture data, protect it and make it secure and in compliance. What has had to come together in order to support what was just described by Missy in terms of the Cerner solution?

Healthcare tech progresses to next level 

Rich Bird

Bird

Bird: Focusing on the outcomes is very important. It delivers confidence to the clinical team, always at the front of mind. But it provides that in a way that is secured, real-time, and available, no matter where the care team are. That’s very, very important. And the fact that all of the devices are connected poses great potential opportunities in terms of the next evolution of healthcare technology.

Until now we have been digitizing the workflows that have always existed. Now, for me, this represents the next evolution of that. It’s taking paper and turning it into digital information. But then how do we get more value from that? Having Wi-Fi connectivity across the whole of a site is not something that’s easy. It’s something that we pride ourselves on making simple for our clients, but a key thing that you mentioned was security around that.

When you have everything speaking to everything else, that also introduces the potential of a bad actor. How do we protect against that, how do we ensure that all of the data is collected, transported, and recorded in a safe way? If a bad actor were to become a part of external network and internal network, how do we identify them and close it down?

Working together with our partners, that’s something that we take great pride in doing. We spoke about mobility, and outside of healthcare, in other industries, mobility usually means people have wide access to things.

devicesBut within hospitals, of course, that mobility is about how clinicians can collect and access the data wherever they are. It’s not just one workstation in a corner that the care team uses every now and again. The technology now for the care team gives them the confidence to know the data they are taking action on is collected correctly, protected correctly, and provided to them in a timely manner.

Gardner: Missy, another part of the foundational technology here is that algorithm. How are machine learning (ML) and AI coming to bear? What is it that allowed you to create that algorithm, and why is that a step further than simple reports or alerts?

Ostendorf: This is the most exciting part of what we’re doing today at Cerner and in healthcare. While the St. John’s Sepsis Algorithm is saving lives in a large-scale way – and it’s getting most of the attention — there are many things we have been able to do around the world.

Deirdre brought up Ireland, and even way back in 2009 one of our clients there, St. James’s Hospital in Dublin, was in the news because they made the decision to take the data and build decision-making questions into the front-end application that the clinicians use to order a CT scan. Unlike other X-rays, CT scans actually provide radiation in a way that’s really not great. So we don’t want to have a patient unnecessarily go through a CT scan. The more they have, the higher their risks go up.

They take the data and build decision-making questions into the front-end of the application the clinicians use to order a CT scan. We don’t want to have a patient unnecessarily go through a CT scan. Now with ML, it can tell the clinician whether the CT scan is necessary for the treatment of that patient.

By implementing three questions, the computer looks at the trends and why the clinicians thought they needed it based on previous patients’ experiences. Did that CT scan make a difference and how they were diagnosed? And now with ML, it can tell the clinician on the front end that, “This really isn’t necessary for what you are looking for to treat this patient.”

Clinicians can always override that, they can always call the x-ray department and say, “Look, here’s why I think this one is different.” But in Ireland they were able to lower the number of CT scans that they had always automatically ordered. So with ML they are changing behaviors and making their community healthier. That’s one example.

Another example of where we are using the data and ML is with the Cerner Opioid Toolkit in the United States (US). We announced that in 2018 to help our healthcare system partners combat the opioid crisis that we’re seeing across America.

Deirdre, you could probably speak to the study as a clinician.

Algorithm assisted opioid-addiction help

Stewart: Yes, indeed. It’s interesting work being done in the US on what they call Opioid-Induced Respiratory Depression (OIRD). It looks like approximately 1 in 200 hospitalized surgical patients can end up with an opioid-induced ventilatory impairment. This results in a large cost in healthcare. In the US alone, it’s estimated in 2011 that it cost $2 billion. And the joint commission has made some recommendations on how the assessment of patients should be personalized.

It’s not just one single standardized form with a score that is generated based on questions that are answered. Instead it looks at the patients’ age, demographics, previous conditions, and any other history with opioid intake in the previous 24 hours. And according to the risks of the patient, it then recommends limiting the number of opioids they are given. They also looked at the patients who ended up in respiratory distress and they found that a drug agent to reverse that distress was being administered too many times and at too high a cost in relation to patient safety.

Now with the algorithm, they have managed to reduce the number of patients who end up in respiratory distress and limit the number of narcotics according to the specific patients. It’s no longer a generalized rule. It looks at specific patients, alerts, and intervenes. I like the way our clients worldwide work in the willingness to share this information across the world. I have been on calls recently where they voiced interest in using this in Europe or the Middle East. So it’s not just one hospital doing this and improving their outcomes — it’s now something that could be looked at and done worldwide. That’s the same whenever our clients devise a particular outcome to improve. We have seen many examples of those around the world.

Ostendorf: It’s not just collecting data, it’s being able to actualize the data. We see how that’s creating not only great experiences for a partner but healthier communities.

Gardner: This is a great example of where we get the best of what people can do with their cognitive abilities and their ability to contextualize and the best of the machines to where they can do automation and orchestration of vast data and analytics. Rich, how do you view this balancing act between attaining the best of what people can do and machines can do? How do these medical use cases demonstrate that potential?

Machines plus, not instead of, people 

Bird: When I think about AI, I grew up in the science fiction depiction where AI is a threat. If it’s not any taking your life, it’s probably going to take your job.

But we want to be clear. We’re not replacing doctors or care teams with this technology. We’re helping them make more informed and better decisions. As Missy said, they are still in control. We are providing data to them in a way that helps them improve the outcomes for their patients and reduce the cost of the care that they deliver.

It’s all about using technology to reduce the amount of time and the amount of money care costs to increase patient outcomes – and also to enhance the clinicians’ professionalism.

Missy also talked about adding a few questions into the workflow. I used to work with a chief technology officer (CTO) of a hospital who often talked about medicine as eminence-based, which is based on the individuals that deliver it. There are numerous and different healthcare systems based on the individuals delivering them. With this digital technology, we can nudge that a little bit. In essence, it says, “Don’t just do what you’ve always done. Let’s examine what you have done and see if we can do that a little bit better.”

We know that personal healthcare data cannot be shared. But when we can show the value of the data when shared in a safe way, the clinical teams can see the value generated . It changes the conversation. It helps people provide better care.

The general topic we’re talking about here is digitization. In this context we’re talking about digitizing the analog human body’s vital signs. Any successful digitization of any industry is driven by the users. So, we see that in the entertainment industry, driven by people choosing Netflix over DVDs from the store, for example.

When we talk about delivering healthcare technology in this context, we know that personal healthcare data cannot be shared. It is the most personal data in the world; we cannot share that. But when we can show the value of data when shared in a safe way — highly regulated but shared in a safe way — the clinical teams can then see the value generated from using the data. It changes the conversation to how much does the technology cost. How much can we save by using this technology?

For me, the really exciting thing about this is technology that helps people provide better care and helps patients be protected while they’re in hospital, and in some cases avoid having to come into the hospital in the first place.

Gardner: Getting back to the sepsis issue as a critical proof-point of life-enhancing and life-saving benefits, Missy, tell us about the scale here. How is this paying huge dividends in terms of saved lives?

Life-saving game changer 

Ostendorf: It really is. The World Health Organization (WHO) statistics from 2018 show that 30 million people worldwide experience a sepsis event. In their classification, six million of those could lead to deaths. In 2018 in the UK, there were 150,000 annual cases, with 44 of those ending in deaths.

You can see why this sepsis algorithm is a game-changer, not just for a specific client, but for everyone around the world. It gives clinicians the information they need in a timely manner so that they can take immediate action — and they can save lives.

doctorRich talked about the resources that we save, the cost that’s driven out, all those things are extremely important. When you are the patient or the patient’s family, that translates into a person who actually gets to go home from the hospital. You can’t put a dollar amount or an efficiency on that.

It’s truly saving lives and that’s just amazing to think that. We’re doing that by simply taking the data that was already being collected, running that through the St. John’s sepsis algorithm and alerting the clinicians so that they can take quick action.

Stewart: It was a profound moment for me after Hamad Medical Corp. in Qatar, where I had run the sepsis algorithm across their hospitals for about 11 months, did the data and they reckoned that they had potentially saved 64 lives.

And at the time when I was reading this, I was standing in a clinic there. I looked out at the clinic, it was a busy clinic, and I reckoned there were 60 to 70 people sitting there. And it just hit me like a bolt of lightning to think that what the sepsis algorithm had done for them could have meant the equivalent of every single person in that room being saved. Or, on the flipside, we could have lost every single person in that room.

Mothers, fathers, husbands, wives, sons, daughters, brothers, sisters — and it just hit me so forcefully and I thought, “Oh, my gosh, we have to keep doing this.” We have to do more and find out all those different additional areas where we can help to make a difference and save lives.

nurseGardner: We have such a compelling rationale for employing these technologies and processes and getting people and AI to work together. In making that precedent we’re also setting up the opportunity to gather more data on a historical basis. As we know, the more data, the more opportunity for analysis. The more analysis, the more opportunity for people to use it and leverage it. We get into a virtuous, positive adoption cycle.

Rich, once we’ve established the ability to gather the data, we get a historical base of that data. Where do we go next? What are some of the opportunities to further save lives, improve patient outcomes, enhance patient experience, and reduce costs? What is the potential roadmap for the future?

Personalization improves patients, policy 

Bird: The exciting thing is, if we can take every piece of medical information about an individual and provide that in a way that the clinical team can see it from one end of the user’s life right up to the present day, we can provide medicine that’s more personalized. So, treating people specifically for the conditions that they have.

Missy was talking about evaluating more precisely whether to send a patient for a certain type of scan. There’s also another side of that. Do we give a patient a certain type of medication?

When we’re in a situation where we have the patient’s whole data profile in front of us, clinical teams can make better decisions. Are they on a certain medication already? Are they allergic to a medication that you might prescribe to them? Will their DNA, the combination of their physiology, the condition that they have, the multiple conditions that they have – then we start to see that better clinical decisions can be made. We can treat people uniquely for the specific conditions.

At Hewlett Packard Labs, I was recently talking with an individual about how big data will revolutionize healthcare. You have certain types of patients with certain conditions in a cohort of patients, but how can we make better decisions on that cohort of patients with those co-conditions? You know, with at a specific time in their life, but then also how do we do that from an individual level of individuals?

Rather than just thinking about patients as cohorts, how could policymakers and governments around the world make decisions based on impacts of preventative care, such as more health maintenance? We can give visibility into that data to make better decisions for populations over long periods of time.

It all sounds very complicated, but my hope is, as we get closer, as the power of computing improves, these insights are going to reveal themselves to the clinical team more so than ever.

There’s also the population health side. Rather than just thinking about patients as individuals, or cohorts of patients, how could policymakers and governments around the world make decisions based on impacts of preventative care, such as incentivizing populations to do more health maintenance? How can we give visibility into that data into the future to make better decisions for populations over the longer period of time?

We want to bring all of this data together in a safe way that protects the security and the anonymity of the patients. It could provide those making clinical decisions about the people that are in front of them, as well as policymakers to look over the whole population, the means to make more informed decisions. We see massive potential around prevention. It could have an impact on how much healthcare costs before the patient actually needs treatment.

It’s all very exciting. I don’t think it’s too far away. All of these data points we are collecting are in their own silos right now. There is still work to do in terms of interoperability, but soon everybody’s data could interact with everybody else’s data. Cerner, for example, is making some great strides around the population health element.

Gardner: Missy, where do you see accelerating benefits happening when we combine edge computing, healthcare requirements, and AI?

At the leading edge of disease prevention

Ostendorf: I honestly believe there are no limits. As we continue to take the data in in places like in northern England, where the healthcare system is on a peninsula, they’re treating the entire population.

Rich spoke to population health management. Well, they’re now able to look across the data and see how something that affects the population, like diabetes, specifically affects that community. Clinicians can work with their patients and treat them, and then work the actual communities to reduce the amount of type 2 diabetes. It reduces the cost of healthcare and reduces morbidity rate.

That’s the next place where AI is going to make a massive impact. It will no longer be just saving a life with the sepsis algorithm running against those patients who are in the hospital. It will change entire communities and how they approach health as a community, as well as how they fund healthcare initiatives. We’ll be able to see more proactive management of health community by community.

hpe-logoGardner: Deirdre, what advice do you give to other practitioners to get them to understand the potential and what it takes to act on that now? What should people in the front lines of caregiving be thinking about on how to best utilize and exploit what can be done now with edge computing and AI services?

Stewart: Everybody should have the most basic analytical questions in their heads at all times. How can I make what I am doing better? How can I make what I am doing easier? How can I leverage the wealth of information that is available from people who have walked in my shoes and looked after patients in the same way as I’m looking after them, whether that’s in the hospital or at home in the community? How do I access that in an easier fashion, and how do I make sure that I can help to make improvements in it?

Access to information at your fingertips means not having to remember everything. It’s having it there, and having suggestions made to me. I’m always going back and reviewing what those results and analytics are to help improve the next time, the next time around.

From bedside to boardroom, everybody should be asking themselves those questions. Have I got access to the information I need? And how can I make things better? What more do I need?

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy. Sponsor: Hewlett Packard Enterprise.

You may also be interested in:

About Dana Gardner

Dana Gardner is president and principal analyst at Interarbor Solutions, an enterprise IT analysis, market research, and consulting firm. Gardner, a leading identifier of software and cloud productivity trends and new IT business growth opportunities, honed his skills and refined his insights as an industry analyst, pundit, and news editor covering the emerging software development and enterprise infrastructure arenas for the last 18 years. Gardner tracks and analyzes a critical set of enterprise software technologies and business development issues: Cloud computing, hybrid IT, software-defined data center, IT productivity, multicloud, AI, ML, and intelligent enterprise. His specific interests include social media, cloud standards and security, as well as integrated marketing technologies and techniques. Gardner is a former senior analyst at Yankee Group and Aberdeen Group, and a former editor-at-large and founding online news editor at InfoWorld. He is a former news editor at IDG News Service, Digital News & Review, and Design News.
This entry was posted in artificial intelligence, big data, Business intelligence, Cloud computing, data analysis, electronic medical records, Enterprise transformation, healthcare, Hewlett Packard Enterprise, Information management, Internet of Things, machine learning, Security and tagged , , , , , , , , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s