Customer Innovations

[1:41] We benefit from public-private partnerships. This is biomed's version of a health information exchange: all of the wheelchairs from the area hospitals accumulating lovingly in the dump, and blessedly, the guys at the dump bring them back to us when enough accumulate. We're very grateful.

[2:04] But at the end of the day, we're really here to enable our patients to enjoy what matters to them – like this lovely art, where I have to say the respect for the institution is palpable in the choice not to tag the wheelchair. That's biomed. We do it with 200 committed clinical engineers and biomedical technicians caring for 100,000 regulated medical devices scattered over 10 million square feet of hospital space.

But as I just showed you, 10 million square feet is just the beginning. The reality is, they’re drifting across eastern Massachusetts because we’ve got doors, ambulance ramps, helicopters, and even a boat dock. It’s a hard job living on the front lines of healthcare, and that hard job requires break-fix. The bed is broken, the call comes in, the tech goes out, and the bed is repaired.

[2:57] I'll let you in on a secret: this ticket got really fast service. Does anyone recognize the left hand in the picture? That was my wife's bed when our second child was born.

All of those medical devices have a routine maintenance schedule. Have you ever driven a little past 5,000 miles on your oil change? It's okay. I forgive you – I do that too. But we never go past dates on our regular inspection of our medical equipment. I think this was our second child's newborn nursery warmer. It had been appropriately inspected.

[3:36] Then there's critical downtime response, where routine maintenance has to give way to the critical incident of the day. Who recognizes what our engineers are huddled around here? It's the blue screen of death from CrowdStrike. The pager goes off at 2 a.m., and that doesn’t change the fact that there’s a heart-liver transplant scheduled for the next day. You’ve got to get up and figure out how to fix it because we're going to do that transplant no matter what. That’s the commitment to the patient.

And when we do that work, we do it over a wide array of technologies. Our most common technologies are our infusion pumps – large volume, micro infusion, patient-controlled – well over 10,000 of them, delivering millions of infusions a year. Our point-of-care imaging and diagnostic modalities – sophisticated, fragile equipment that used to be limited to diagnostic suites and operating rooms – is now rolled down the hallways to the hard-charging emergency department.

[4:40] In our operating rooms: hybrid ORs, fixed imaging modalities, surgical integration, surgical navigation, and robotic-assisted surgery. This is the stuff of prime-time dreams and everyday miracles. Nearest and dearest to my heart is the intensive care unit.

You call it an intensive care unit, and it sounds like a place, a building, a room. That’s not quite right. It’s actually a scaffolding on which committed doctors, nurses, respiratory therapists, and allied health professionals build a one-of-a-kind machine designed to get you through the single most vulnerable day of your life. It comes into existence for you and goes away the minute you’re discharged.

They build that machine on the framework of the ICU out of the Lego bricks we provide them: critical care ventilators for lungs, mechanical circulatory supports for hearts, renal replacement therapy for kidneys, ventilators, tens of channels of infusion – a one-of-a-kind machine. And our hope is for one of these cases to provide perfect maintenance and ergonomics that allow us to never be the main character. We are at the end of the day, first and foremost, seller dwellers. We want to stay in that basement wonderland of the hospital and never be noticed. That's perfect success.

[5:58] Unfortunately, as the ICU and operating room got more technologically dense, we started to become the main character. Connecting all this medical equipment and getting the data from that medical equipment – the vital signs – into our electronic medical record was becoming a full-time job for nurses, respiratory therapists, and anesthesiologists.

We can all get fast on a 10-key, but that’s not a real integration engine – and you all know that’s just a computer. Moving that handful of numbers from our specialized medical device computers into the medical record, it seemed obvious there had to be a better way than doing it by hand

[6:45] This is the birth of biomedical device integration. It’s a really simple concept. You walk into a doctor's office, and you get plugged into the machine. That’s the bad part that you’re not excited about. But the point is, once you're plugged in, we can plug the machine into the medical record, and do that documentation automatically. That’s the entire concept: move the number that’s about you from the machine you’re already plugged into, into the medical record using a computer instead of a skilled professional's fingers working a keypad.

The devil’s always in the details. There are about as many medical device integration patterns as there are medical device fingers. But this is the most common one: the aggregation server, where all the devices of a given kind phone home to a specific aggregation engine that’s designed to collect the data from that device, papering over its proprietary format and emitting HL7 messages. Those HL7 messages flow forward to our InterSystems message-passing middleware before going on to our medical record

[7:55] And this is how the system operated until COVID, and we were happy with it. We never thought about that message passing middleware. It was the perfect technology for biomed, unnoticed. But then COVID hit, and suddenly we were asked to do more. We needed to turn on secondary web viewers, so our staff could cover more patients from farther away. And we needed to route alarms to the nurse's iPhone so that we could quiet down our double-capacity ICUs. That shifted attention to middleware and how we could get more out of it.

[8:34] Where are we starting from? We’re starting from the electronic medical record, which is always growing but still has a limited number of data points. A few vital signs are collected every four hours on med-surg floors, every minute under general anesthesia, and maybe every 10 seconds if deeply anesthetized. That’s what’s growing in our medical record – it keeps accumulating over time.

Meanwhile, under our aggregation server is where we spend most of our time. It works like a black box recorder in a modern aircraft. It captures every single bit flowing from our medical devices and is available for full-disclosure investigations if something goes wrong. But just like a black box, the next successful episode of care will overwrite the oldest one. If we don’t go back to investigate a problem, it’s gone forever. That’s where we’re starting from.

[9:26] This is COVID, when we’re just getting used to the idea that maybe we can get more out of that message passing middleware. That’s our work. But I told you I had a problem. My problem is your problem. It’s the problem of medicine writ large. Many of you probably got into this during the American Reinvestment and Recovery Act incentive program. You’re used to the idea of the triple aim, which has been expanded to the quadruple aim. These are the goals American healthcare has set for itself to improve

[9:55] What do we want? We want it to cost less. We want to burn our doctors and nurses out more slowly – or better yet, not at all. We want better population health outcomes. We want it to be better. And we want to provide a better patient experience. These are all very reasonable goals that we’ve committed to as an industry.

[10:22] What does biomed have to do with this, though? I told you we live in the cellar, providing lights-on support for medical devices. It feels like we’re part of that expensive problem, right? The truth is, we grew out of this bottom half. We grew out of a need to provide better population health. We started from a patient safety perspective.

[10:49] This is the famous To Err is Human report from the Institute of Medicine, which popularized the idea that many Americans die from preventable medical errors. More recently, a BMJ article made headlines suggesting that preventable medical errors could be the third most common cause of death in the US. Those are scary numbers. You don’t need to be scared next time you go to the hospital, though.

Each study in its own time raised methodological questions, which lowered the rate of preventability below 10%. That brings it to only about 7,000 people a year who die from preventable medical errors. That feels better than a quarter of a million. On the other hand, it’s still far too many. Every single one is a tragedy, and we feel them acutely

[11:40] These are the events that take us back into that aggregation server to do a full disclosure investigation. The events that send us looking for a root cause. The events that remind us where clinical engineering came from 50 years ago: the need to provide safety, to protect patients from electrical injuries and fires in operating rooms.

Those problems remain today, but they’re blessedly rare. We think fewer than 100 people will be caught on fire in an operating room in the US this year. That’s better than it used to be, and still far too many. We’re desperate. We were born to provide safer care.

[12:23] What about the top part? American medicine is about 20% of GDP. Everybody agrees it’s too expensive. And COVID showed us that you can burn out a healthcare professional who takes 15 years to train in just 15 days. That’s no way to run a system, and it’s going to have real consequences.

[12:40] These are Department of Health and Human Services Labor Statistics. Years on the x-axis, going out to 2036. Staffing adequacy on the y-axis. Unfortunately, zero is at the top of the slide. We don’t have enough. We don’t have enough registered nurses today, and the problem’s only going to get worse over the next decade.

Although we have enough anesthesiologists today, we’re going to be short 6,000 of them in the next decade. Why nurses and anesthesiologists? Because they’re our most common customers as clinical engineers. I could have shown you many other medical professionals. This is the nature of medicine in the decade ahead.

[13:24] We aspire to provide safer, cheaper, more effective care, but we’re going to have to do it with fewer of the skilled professionals that define the system today. And it’s not just the skilled professionals. This is the OECD’s estimate of old-age dependency. Every line is a country; the US is the red line. Years on the x-axis, old-age dependency on the y-axis. What’s old-age dependency? If you’ve ever been in a hospital, you know the oldest person is probably in the bed, with a lot of prime-age workers standing around. This shows how many prime-age workers you have for every older adult.

[14:05] When we built modern acute care medicine, we had four or five prime-age workers per older adult. As we get into the century ahead, we’ll have two. We’ve got to go from five to two, with 50 years to do it – all while figuring out how to be cheaper and safer. The good news: we don’t have to invent; we just have to copy.

[14:36] When regularly scheduled commercial air travel started in 1938 with the Panama Clipper service, there was a flight crew of five. It looked like the bridge of a ship: helmsman, captain, navigator, engineer, radio operator. When you flew down here to Orlando this week, how many of you saw five people at the front of the plane? Not one of you. I bet you saw two people. By 1967, two years before man landed on the moon, the first two-person crew was introduced. They did it – five to two in half a century. That’s exactly what we have to do.

[15:31] How do you do it? The exact same way they did: automation and routinization. Complexity is still there; you have to shift it off the people. We’re going to do that with closed-loop medical devices. This terrifies people at a biomed meeting. You take the device, transduce the signal, pass it into a software control system, and directly modify the treatment delivered to the patient. It’s an autopilot. Full authority digital engine control. Scares everyone. And yes, all the things that can go wrong – they are all going to happen. We’ll have to get through it together. When we do, we hope for what aviation got: something safer, more reliable, and more affordable. It’s not an aspiration – it’s a necessity. We’re going from five to two.

[16:28] What’s the data that will take us there? This is what flows out of our medical devices today: waveforms, trends, alarms. You want them wiggly, not flat. Trends count occurrences – heart rate, for example, this person had 67 in the prior minute. And then there are alarms. Every device enunciates its own alarms as HL7 messages. OBX5 – that’s the waveform. That’s the picture doctors, nurses, anesthesiologists, and respiratory therapists use to guide lifesaving ICU treatment.

[17:30] The challenge: we throw all of this away. The ring buffer in the aggregation server gets overwritten every 90 days. To achieve automated medical devices safely, we need this data to be the basis of in-silico trials to evaluate devices before deploying them in hospitals. If we want that, we must stop throwing this data away today. That’s the goal.

That was the problem I took to InterSystems. How can I stop throwing this data away today? Because it's going to be the only tool I have to safely get to the future. And the answer is good news: you’re already sending HL7 traffic into performant message-passing middleware. We’re quite performant. Even at full outflow, it handles the load on hardware resembling my laptop more than mainframes.

[18:37] You might say, “Wait a second – this is a weird data structure.” Some of those devices run at 500 Hz, others at 0.5 Hz. The timelines aren’t synchronized. There’s no end. It’s not a document; it flows forever. Some ICU encounters last a day, others 100 days. Where do you put all that? Well, of course, the answer is – it’s a good thing we’ve got a really flexible persistence engine under the covers. Why not put it there?

We were also fortunate to benefit from a fabulous partnership with InterSystems over the past three years, building a data structure that can accommodate this idiosyncratic data type and store it in that persistence layer.

[19:14] But then they went further and said, “Well, if you’re going to make this innovation impactful, you’re going to have to expose that data to the workforce you have. The workforce we have – they’re good at using SQL. So, what if we made that data accessible to you through a SQL projection so your workforce can be productive today?” And we said, “That’s fabulous.”

And so, that’s exactly what we did. We started leveraging the middleware more during COVID to give our staff the better ergonomics they required to meet the demands of that moment. And we’re starting to capture that data and deploy it forward in this improved SQL projection so that we can build the tools we know we need for the century ahead.

[19:58] I regret that I’ve come all the way from a medical school and didn’t tell you about the Krebs cycle. That was a missed opportunity on my part. I hope you’ll forgive me. Instead, what I did was tell you what clinical engineers do. We take care of all of the devices in the hospital that make a hospital a hospital instead of a Motel 6.

I told you about what the century ahead holds: an unending aspiration that our care should be safer, despite the fact that we’re going to provide it to our patients with half as many workers per patient. And I told you about how InterSystems is helping us get there by storing our HL7 traffic in real time in a purpose-built data structure that only they could provide through a partnership that only they were willing to offer.

[20:42] So with that, I want to say thank you so much for your attention, and I can’t wait to see what amazing things you guys will do with InterSystems’ technology to help solve the problems that are coming for healthcare in the century ahead, no matter what.