Using the 5Ds of IoT for medical devices
I’m a fan of frameworks or techniques that help people understand our complex environment and reduce risk.
Using the 5Ds of IoT for medical devices
For example, there’s Michael Porter’s Five Forces model, Failure Mode and Effects Analyses, Monte Carlo simulations and many others that facilitate decision making. As the acronym IoT (Internet of Things) becomes as ubiquitous as WTF, it behooves any professional working with connected medical devices to get acquainted with an evolving landscape that can easily derail any honorable effort to improve patient outcomes.
I’ve found the 5Ds of IoT to be a useful model in classifying the myriad of topics and questions that need to be addressed to succeed in this space. They can also serve as a guidepost for smart industrial and consumer products. The 5Ds are Decision, Devices, Data, Design and Deployment.
Decision
“What outcome is my company seeking to improve with this new technology?” should be the starting block for any connected device development endeavor.
At the recent BioMEDevice conference in San Jose, Dan Pettus, BD/Carefusion VP of Connectivity and IT, discussed how his company came to embrace connectivity for their CareFusion Alaris PC pump after recognizing the enormity of issues affecting the delivery of infusion therapy. For example, it typically takes over 30 steps to get medication to a patient. Infusion-associated adverse drug events add more than $2 billion to annual healthcare costs. And 93 percent of nursing staff said they strongly agreed about the importance of having the right data in patient management. Those are pretty compelling reasons to move in a new direction. When considering connecting a medical device, you need to start with the size of the problem you’re trying to solve and weigh the trade-offs.
“Connected health is not for the faint of heart,” Pettus said.
Thanks to the interoperability of their Alaris pump with Electronic Health Records (EHR), the nurse has to only scan the bar code on the patient’s wrist then scan the Alaris pump. The pump automatically programs the settings, which saves about 30 keystrokes. More importantly, it reduces data entry errors. In addition, because insurers are demanding more accurate data for reimbursement, a 400-bed hospital that can produce evidence of when procedures start and stop can recoup about $1 million.
While figuring out the benefit may not be as straightforward, it’s imperative that organizations spend time collecting evidence — costs, time, errors, steps, etc. — so they have a benchmark to work against. In her BioMEDevice talk, Stephanie Kute, Manager of Advanced Analytics and Health Research at Batelle, offered a few standard metrics worth considering:
1) AHRQ Quality Indicators, which are provided by the Department of Health and Human Services and regularly used for reimbursement practices, include:
- Hospital-based metrics such as inpatient mortality and utilization of procedures
- Patient safety indicators (PSI) that focus on potentially avoidable complications and errors such as pressure ulcers or foreign objects left in patients after surgery
- Outpatient-based metrics, such as Prevention Quality Indicators for chronic diseases like diabetes, heart failure and asthma that can help providers keep patients out of the hospital.
2) Patient Experience and Engagement Metrics
- Patient experience is collected by a survey known as HCAHPS (Hospital Consumer Assessment of Healthcare Provider and Systems)
- The Patient Activation Measure (PAM) measures patient engagement and education.
3) Cost of Care
- The National Quality Forum endorses the PMPM index, which would be the total cost of care, population-based, per member per month. An example would be hospital-level, risk-standardized payment associated with a 30-day episode-of-care for heart failure.
The decision to pursue a connected device project has to be strategically supported form the top to succeed.
Devices
As mentioned above, interoperability is a critical aspect in this landscape and it refers to how your devices play with other devices. According to George Panagiotopoulos, Senior Clinical Tech Manager at Kaiser Permanente, there are about “700 different types of medical devices in a hospital used at any point in time.” Since your device is likely to depend on another to deliver data at the right time for the right patient, the synchronization of connectivity cannot fail. So what interoperability requirements should you consider? Here are some requirements offered by Pettus of BD/Carefusion:
- Cyber-security. Security is a journey, not a one-time event. You need constantly monitor security issue to stay on top of them.
- Fast connect and roaming. Most hospitals are in 802.11, which is “great for Starbucks,” but since these are mobile devices they have to connect during transportation — down hallways, up elevators, in different rooms, etc.
- Store and forward. Given the above, your device may not necessarily need to be optimized for synch capability. ‘Store and forward’ may be sufficient.
- Availability and integration. The Integrating the Healthcare Enterprise – Patient Care Device (PCD) domain addresses the integration of medical devices into the healthcare enterprise such that information from vital signs, physiological monitors, ventilators, infusion pumps, and anesthesia workstations could be exchanged and alerts could be managed. See this link for more info.
Also, let’s not forget that medical devices are heavily regulated, not only by FDA regulations but also by HIPAA and BAA requirements. This brings a host of other considerations such as encryption, hosting capabilities and patch management. So finding the right partner to help you navigate those intricacies is important.
Data
At the end of the day, it’s the data that will inform your progress toward better patient outcomes. Having a good understanding of the type, access, processing, analytics, storage, security and privacy of data is critical.
The key challenge, however, is how to effectively use data to motivate patients to adhere to their treatments when there are countless barriers — ease of use, cost, procrastination — that get in the way.
One of the companies aiming to demonstrate the value of digital health data is Evidation Health. Evidation began as a collaboration between GE Ventures and Stanford Health Care that focused on demonstrating better health outcomes with digital health solutions and new approaches to predictive analytics. Evidation’s CEO, Deborah Kilpatrick and her colleagues are using machine learning and behavioral economics to transform the way consumers are incentivized to be more proactive in managing their own health.
At Design Concepts, we have helped clients design new products that, with data transparency, are demonstrating improved compliance compared with older products. For example, Bioventus’ EXOGEN, the only FDA-approved bone-healing device that uses ultrasound, was able to improve patience compliance by 50 percent through the use of data shared with patients and doctors.
While most companies are very guarded as to what they have found, being intentional about what metric you are trying to improve can yield new discoveries. More often than not, though, organizations will find themselves with an abundance of data and then run algorithms to see what insights can be unpacked. In the world of data science, you either start with a hypothesis and use data to prove/disprove it or you start with big data and see what correlations you can draw. Most likely, your organization will use both techniques to discover new patterns.
Data provenance is a term that was often used during the conference. It refers to the origin, lineage or source of data. If a device can send meta-data (i.e.. how and when the data was acquired, device usage, etc.), then issues can be resolved faster by pinpointing where the problem might have occurred. This is especially important during recalls, as a company no longer has to do a massive recall but simply identify those devices that exhibit similar properties.
Finally, as the data collection journey begins, standardization is critical, especially if your device operates in regions where data can be logged in Fahrenheit or Celsius. The last thing you’ll want is to have abnormal readings that will trigger alerts that are unnecessary simply because temperature readings across countries were not normalized.
Design
The sweet spot of design comes down to the intersection between desirability, feasibility and viability. During a panel session titled “How to Understand What Hospitals, Physicians, and Patients Want from Connected Devices,” Ben Rosner, Chief Medical Officer at Healthloop, said design is quite important given that healthcare companies are in “fortress medicine” mode, which is very slow moving, and it’s hard to change patients’ behavioral patterns. So the challenge is “how do you have minimal disruption in the workflow?” That’s the question that designers are constantly trying to answer. You can certainly have a great device, but if it doesn’t fit well into the larger ecosystem it won’t live for long.
When there are 700 medical devices being used in a hospital at any given time, we need to think about human factors (ergonomic and cognitive) and how clinicians and patients interact with devices. Does the design evoke trust, sophistication and reliability? In the age of beautiful smart phones, there’s a high standard for medical devices to follow. Operational reliability is almost a foregone conclusion — now we need devices that speak to our senses.
One of the challenges is how to make a device new without increasing the learning curve. Panelist Brian Birch of AIG offered an analogy worth repeating: “When you’re designing something, it’s like getting into a new car that you’ve never driven before — it should be familiar and not too difficult to learn.”
During a session titled “Cloud DX — The Gold Standard in Advanced Health Monitoring Solutions & Medical Grade Wearables,” Robert Kaul, CEO of Cloud DX, offered an example that the number-one reason why people stop using pulse oximeters — the AAA battery dies and procrastination in replacing it leads to lack of use. This seems like a rather simple problem, but when medical device manufacturers are also thinking about how a consumer would use this at home they have to account for various issues that might get in the way of adherence.
Cloud DX has gone to great lengths to make their Connected Health Kit as inexpensive, accessible and data-rich as possible in order to bring home Qualcomm’s Tricorder X-Prize of $10 million. As stated on X-Prize website, “The winning team will develop a Tricorder device that will accurately diagnose 13 health conditions (12 diseases and the absence of conditions) and capture five real-time health vital signs, independent of a health care worker or facility, and in a way that provides a compelling consumer experience.” So, as one of the handful of finalist in this competition that started in June 2013, Cloud DX is taking design very seriously.
Deployment
The last “D” is the one that brings it all together to create a sustainable product.
Given all the complexities in bringing a connected device to market, the expression "eat an elephant one bite at a time" could never be truer. Ultimately, your organization wants to build and launch a product that is unique and successful in the marketplace. In one of the last panels of the conference, “How to Set Your Connected Health Solutions Apart,” representatives from Livongo, Zing Health, iRhythm and Stryker Medical discussed ways to deploy and differentiate your product.
Theo Than of Zing Health, a startup that helps families navigate the cancer treatment process, used an “identify, invent and implement” approach that was needs-based with full clinical immersion and market research. His company spent months with pharmacists, case managers, assisted living entities and patients at home to learn the entire health experience. They discovered that while seniors did not understand the concept of Uber (using an app to call a cab-like service) they did like the convenience of GoGoGrandparent.com, which basically uses a similar model but relies on an actual call with a person. This speaks to the importance of removing fears and unknowns in a process that is otherwise already very effective with some users. Zing Health hones in on the pivotal points that create opportunities for behavioral change. For example, if patients said they stopped smoking when they were admitted for a stroke, that’s when they were also more receptive to giving new products a chance.
Patty Siller of Livongo, a company that provides coaching tools to employers so their employees can manage diabetes, spoke about the importance of creating a Minimal Viable Product (MVP) to quickly learn how users interact with a product and come up with key features that signal you’re going in the right direction. In two months, they did a survey, talked with consumers, and developed an app to share with a subset of the population to modify and iterate. Livongo understood that consumers need to have a delightful experience when using a new product so they looked at what diabetes supplies they might need and proactively sent them to the patients’ homes before they knew they needed them. Finally, because Livongo knows they’re competing for all of the consumer’s time, they use models that work in other industries — gamification, challenges, goals, badges and communities.
Lives can be at stake when you are developing a medical device, so make sure your organization considers the 5Ds of IoT as it looks to improve patient outcomes and be profitable along the way.
Written by Edgar Hernandez