Be Prepared for Accelerated Adoption of AI in Wearables

By Maria Fontanazza
November 29, 2019


• Categories: 04-Product Design & Development, Business Analysis, RC - Device Design & Manufacturing

The function of healthcare technology in the home will no doubt continue to increase as patients seek more convenience and control over their health, and hospitals pursue cost savings by keeping patients out of hospitals. The adoption of telemedicine and wearables with greater capabilities such as artificial intelligence (AI) will also assume a larger role. In a Q&A with MedTech Intelligence, AI Evangelist Juggy Jagannathan, Ph.D., director of research for 3M Modal, explains where he sees the future of artificial intelligence within medtech in the context of the home healthcare setting.

MedTech Intelligence: What are the current areas of convergence between AI and the medtech industry?

Juggy Jagannathan, Ph.D.: Fundamentally, the question is how do you provide better care for patients at home in a cost effective manner? How do you make patients and clinicians happy? Clinicians are beginning to be bombarded by a tsunami of data that is flooding the medical enterprise, including all of the data from home health sensors. How can clinicians make sense of it and use it? That’s where AI and predictive analytics comes in: It crunches the data and provides predictions that lead to actionable intelligence. Another big area is monitoring and diagnostics. You’re continuously monitoring for something, so you can diagnose various conditions. A classic example is a watch that can monitor the heartbeat and detect AF [atrial fibrillation]. And then there are the health and wellness tracking, fitness, and personalized coaching apps—there’s a lot of AI behind all of it.

There’s another class of therapeutic devices that also use AI. For instance, in hearing aids there is AI technology that enhances the audio so that if you’re in a multiple-person conversation, it will filter and enhance the user experience by allowing the user to focus on one thing. There are also visual aids that help blind people. And last but not least, there are remote monitoring solutions such as remote stethoscopes, which allow the physician at the other end of a telemedicine call to actually examine the patient.

MTI: What applications offer the most promise in implementing artificial intelligence into medical devices used in the home healthcare setting?

Jagannathan: In the macro healthcare environment, there is immense pressure to reduce healthcare costs while ensuring patient care doesn’t suffer.

CMS is rolling out value-based care, and it essentially means that they are providing money to accountable care organizations (ACOs) to take care of a population of patients. This requires ACOs to look at their population of patients and determine their risk profile for various conditions, and how this will impact the bottom line. ACOs can then understand how much money [is needed] to take care of the patient (a quality map). That, in turn, is the money they give to the ACO. So that is the case for the Medicare population.

The same situation occurs across the board for payers. How does a payer manage this process? It falls under the heading of population health management. Payers stratify the population into risk segments—i.e., the segment of the population that requires a significant amount of care, and the healthy people who require less attention. The best way to manage population health is to make sure that people don’t fall sick by promoting healthy living and health awareness. Encouraging people to use health and fitness apps and monitor how well they are doing is one way manage a segment of the population. It is population health management on an unusual scale.

Population health and care coordination is a big task and it takes a significant amount of AI to figure out when to take action. You need to have actionable intelligence about when to take care of a particular patient. Here’s a classic example that recently happened with Apple Watch: Mayo clinic reported on a woman using an Apple Watch, who noticed it said she had atrial fibrillation. She was worried and went to the ER. Sure enough, she was having atrial fibrillation caused by a thyroid problem. She was treated for the thyroid problem and the atrial fibrillation went away. This is a classic example of how a device that is worn at home provides actionable intelligence to the patient. But if all the data is being streamed into an accountable care organization—an organization that is taking care of a population—and there are care coordinators who get the data, then they can react to it and take care of the patients. That’s where all these things have to fit in, because ultimately, you need to take care of a population and ensure they get the right care at the right time so that the best outcomes are achieved at the lowest possible cost.

There’s also the follow up care. This care can happen using tools that monitor the heart rate, the EKG, and other health indicators. These used to be measured in the hospital, but now can be deployed at home. Hospitalization costs are reduced, and the patient is still getting the follow up care. You can monitor these measures remotely just as effectively as at the hospital. That’s another area where monitoring and data collection can help.

Wearables: The Small Parts Drive Innovation | Listen to this complimentary web seminar on-demandTelevisits are becoming more common, and not only in rural areas. In rural areas, it’s common when there isn’t access to a physician. Thirty years ago, telemedicine was also pioneered in prison populations. Today, I think it’s going more mainstream. Imagine you can see a doctor from your home and don’t have to deal with crazy traffic jams or sit in the office for hours because the doctor is way behind. Instead you have your visit via a telecall at home, and then you go to a retail clinic to get your lab work done. You can manage the process more efficiently, and the end result is that you have a happier patient without the time commitments required to see a doctor for 10 minutes. That’s how telehealth and televisits are going to change health care.

This is all in the context of how you take care of the patient at home. We don’t have a system of “health care,” but rather a system of “sick care”— we care for people when they’re sick, instead of helping them stay healthy. These technologies will hopefully flip this paradigm, so the focus is on prevention and health, and ensuring people are doing the right thing for their own health.

MTI: What are the future targets of using AI in medical technology in a home care setting?

Jagannathan: There are hundreds of companies flourishing in this space and focusing on all kinds of sensors. I read an article that in England they’re looking at wearable oral electronics—dentures that you put in your mouth or implants on your teeth that measure your salt and glucose intake and number of bites you take to figure out how much you’re eating. That’s one extreme example of what we can do in terms of completely monitoring all intakes.

There is also a [technology] that will probe your gut to figure out the composition of your biome so you can tailor what you eat to what your biology will be able to digest. We’re talking about all kinds of sensors, and every sensor can ultimately serve as a trigger for actionable intelligence. Ultimately, all these things will lead to a gradual personalization of medicine and help with chronic condition management (i.e., diabetes, hypertension, depression). There’s an explosion of sensors, but the sensors themselves aren’t going to solve anything—they have to measure something, make sense and create actionable insights.

MTI: What types of companies are developing these kinds of sensors? Are there companies that are already working in the healthcare space?

Jagannathan: Many technology companies are trying to get into the healthcare space with applications that are currently outside healthcare. There are sensors embedded in clothing; with smart watches you can do all sorts of things and audio and visual [sensors] can be used for different things. For example, just monitoring speech patterns can tell you if someone is having a stroke. You can even sense a person’s mood based on speech patterns and the way they look. It all sounds a bit invasive, but these technologies can have good uses in the proper context if designed correctly and with patient privacy protection in mind. Someone has to make sense of all the data, but we’re far away from that step and from devising appropriate action based on the data collected.

MTI: What does the future hold for this space?

Jagannathan: I think we’re living in really exciting times. Personalized medicine is in our future—combining an individual’s social, economic and genetic markers, and integrating them to come up with ways to take care of each person in their own, unique way. That’s where we’re headed but we still have a long way to go. There’s a lot of AI tech out there, and AI has to evolve significantly as well—even more than it has evolved up until now. I don’t see any slow down. In fact, the adoption of these technologies is going to accelerate, not slow down.