How IoT will enable continuous care

Hi, I'm Steve Graham, a senior distinguished engineer within the advanced technology collaborative at Optum. Thanks in advance for joining me today for this look at how a quickly evolving technology, the internet of things, or IoT can have a tremendous impact on the healthcare system.

I'm going to be talking about how a Cambrian explosion of new healthcare-oriented devices, wellness devices, smartwatch-based activity monitors, personally worn medical devices, sensors in the home, even smart phone-based digital therapeutics will transform healthcare.

It's not these devices per se that will have an impact on healthcare, but rather the transformative thing is the way we capture the data from these devices, integrate the data into the broader healthcare system, generate patient insights by analyzing this data, and using those patient insights to more effectively engage individuals with their healthcare choices.

We describe a new model of healthcare, we call it 'Continuous care', that uses data about an individual generated by these IoT devices as the way to make healthcare more continuous, anticipatory, and personalized. Let's take a look at this new model. You may not be familiar with the internet of things space, so let me start with a few broad descriptions, and define some acronyms you'll be hearing.

IoT devices capture, and transmit information wirelessly, and enable many modern conveniences. With IoT connected devices, for example, you could tell this virtual assistant on your smart speaker at home to set the temperature to 72 degrees, and turn off the lights on the first floor, or for example, you might check the refrigerator to see if you need to buy milk while you're at the grocery store. While consumer-centric applications have received most of the hype in IoT, there's a huge potential for IoT within healthcare.

When you think about the intersection of IoT, and healthcare, you can consider several possible scenarios. One scenario is the application of IoT in the hospital. For example, using radio frequency identification, or RFID to track large mobile medical equipment. Another scenario is the application of IoT in clinics. For example, using motion detection to determine if a clinical examination room is occupied, to track staff, and patient movement throughout the clinic, et cetera.

A final category relates to personally worn devices, and that's where we'll be spending most of the time today. Fitness trackers like the smartwatches made by Fitbit, Garmin, Apple, and many others, and personally worn medical devices such as continuous blood glucose monitors, network connected asthma inhalers, and the like, offer brand new ways to learn about a patient's health, and wellbeing.

The data that these devices allow us to measure is called 'Patient-generated health data', or PGHD. When we incorporate PGHD with more traditional healthcare data, we can form a more complete picture of the circumstances affecting an individual's health, which can then lead to new opportunities to manage, and sustain it, ultimately allowing for better outcomes, and lower costs.

Let's take a look at how one patient's life might be affected by this new technology, and the PGHD we can capture from it. This is a vignette illustrating a future scenario describing the art of the possible associated with patient-generated health data. Meet Kirsten. Kirsten is an urban professional in her late 30's. Since childhood, Kirsten has struggled with asthma, and when the weather is bad, her asthma can be quite debilitating. Like many people who suffer from asthma, Kirsten struggles to remain compliant with her inhaler medications.

In fact, the last time Kirsten was admitted to the hospital, because of a crisis with her asthma, her doctor prescribed her with an IoT inhaler that tracks when she takes her medications. The doctor also prescribed a network connected spirometer to monitor her lung function. Like many people suffering from a chronic condition, Kirsten has more than one condition.

Kirsten's asthma exacerbates her depression. When she can't breathe properly, she doesn't want to leave the house, increasing her feeling of isolation. Kirsten's depression is managed by an EEG headset, an associated digital cognitive behavioral therapeutic. It's also managed with a smartphone-based application that monitors her depression.

Although the devices come with some condition specific analytics, Kirsten's healthcare team believes that by monitoring her asthma, and depression together, they can more effectively prevent Kirsten from going into crisis, and keep her out of the emergency department.

As a result, a cross morbidity, artificial intelligence analytic was developed to combine pollution data from Kirsten's neighborhood, weather data, a measurement of Kirsten's depression from her EEG, a depression monitoring digital therapeutic, and how compliant Kirsten has been with her inhaler meds, and the status of her lung function from her network connected spirometer.

The analytic generates patient insights to help understand trends in how Kirsten's mental health, and lung health interact with each other, and provide a more complete, timely, and personalized picture of where Kirsten is in her health journey.

One Monday morning, a report on pollution levels in Kirsten's neighborhood was correlated with a weather report for Kirsten's city. This combination suggested a very challenging environment for asthma sufferers. Kirsten's mental health related patient-generated health data from last night also indicated that her current stressful situation at work is taking a toll mentally.

Kirsten's depression might be triggered by adverse effects of her asthma. The cross morbidity, artificial intelligence analytic personalized recursion determines it would be better for Kirsten to take a dose of her inhaler before heading out for her morning commute.

This patient insight then triggers a chat bot to reach out to Kirsten, and emphasize the importance of taking a dose from her inhaler. Kirsten sees the text, and uses her inhaler. A subsequent text then reminds Kirsten to use one of the mindfulness digital therapeutics associated with her EEG device to prepare her for the day.

It is key that the health system interact with Kirsten in a timely, and actionable fashion to help her modify her behavior, and make healthier choices. Because of the timely intervention, Kirsten adjusts her behavior, takes a dose from her inhaler before going out of the door of her apartment. As a result of this change in behavior, Kirsten avoids crisis with her asthma, and is therefore not a guest of the local hospitals emergency department this day.

By avoiding crisis, she also avoids triggering her depression. This means fewer missed days at work for Kirsten. Kirsten's Primary Care Provider, Dr Jessica, receives the data stream about Kirsten, in fact, a data stream of about all her patients who have IoT health devices. Of course, Dr. Jessica doesn't look at the ones in CRS herself, she needs support to securely store, and interpret the data.

In addition to providing advice, and support about how to treat patients like Kirsten, this service also includes a heat map of all of Dr. Jessica's patients that incorporate patient-generated health data, and indicate which of Dr. Jessica's patients would most benefit from timely intervention from the healthcare system. This is an example of the new approaches to healthcare inspired by patient-generated health data.

These changes won't disrupt the healthcare system, but rather by integrating the patient- generated health data, it improves the healthcare system. Kirsten's devices, and digital therapeutics help keep Kirsten's asthma, and depression under control, and not dominating her life, keep Kirsten out of crisis, and out of the emergency department, and help Kirsten's care providers monitor Kirsten, and others like her without being overwhelmed by the data.

Everybody benefits from the healthcare improvements resulting from incorporating IoT, and patient generated-health data into the healthcare system. Kirsten is happy, because it's easier for her to stay compliant with her asthma medication, and she can avoid crisis, thereby avoiding the emergency department, and avoiding time away from work.

Dr Jessica's happy, because she can better help patients like Kirsten be more engaged in their health. Dr Jessica's practice is better, because it can meet it's value-based care objectives, for example, avoiding unnecessary emergency department, and hospital visits, and for Medicare, and Medicaid patients, her office can bill for incorporating patient- generated health data into patient care.

Kirsten's health plan is happy, because patients are more engaged with their healthcare, and fewer patients are admitted into the emergency department. Device manufacturers, and digital therapeutics companies are happy, because their products have been incorporated into the healthcare system. Let me take a moment to clarify an important point. Just as the benefits generated from IoT devices, and patient-generated health data won't accrue to any single stakeholder in healthcare, the opportunity to make a difference in this space isn't limited to device manufacturers, or infrastructure providers.

The real game changer here is the patient-generated health data, and what we could do with that data to deliver more effective health care. A recent study by Deloitte anticipates annual growth of 35 % for solutions, and services that leverage patient generated health data, from $17.1 billion in 2017, to a staggering $77.4 billion in 2022.

There will be opportunities in the market for providers in health plans that focus on population health management, health care operations, care delivery, and consumer experience to design solutions that interpret the patient-generated health data these devices generate in order to solve specific problems across healthcare.

Technologies like this don't develop in a vacuum. Let's take a look at some of the wider trends influencing the way IoT devices in patient-generated health data impact health care. There are some important healthcare trends to consider, consumerization for example. With higher deductible plans becoming more the norm, healthcare is becoming a much more of a consumer good.

This, together with more health-related wearables, and medical device manufacturers becoming more design-oriented, consumer choice is becoming a huge aspect of healthcare. With continuous care, we can look at delivering healthcare that is much more engaging for the individual. The shift to value-based care is also another trend. As more risk is shifted to accountable care organizations, and providers, everyone in the healthcare industry is looking for ways to manage care in a way that avoids big costs.

With continuous care, we can provide patients, providers, and payers with the tools to better manage health, specifically, but not limited to, better chronic care management, and therefore lowering the overall cost of healthcare. Personalization is also a big trend. By using wellness, and medical grade IoT devices, and the associated analytics on the data produced by those devices, we can look for patterns in each individual's health leading to individualized care plans.

By approaching continuous care in a non-siloed way, we can look for patterns across conditions, and treat the person as a whole, and not simply focus on the set of conditions affecting the individual in isolation. Consumer electronic trends are also worth considering. Many more startups, and large firms are innovating in the wellness, and personally worn medical device space.

Wearables like Fitbit, and Apple Watch, and many others are transitioning from just the wellness, and activity monitoring space, to incorporate the ability to capture medical grade data. Smartphones are broadly deployed, and increasingly capable of acting as a personal data hub to interface between the personal health devices, and the healthcare system via Wifi or 5G broadband networks.

Smartwatches are playing a growing role in healthcare, acting as a personal area networking device hub. Smart speakers, and virtual assistants available in one in five homes in the US act as a convenient conduit for delivering health information, and digital therapeutics.

We also should think about artificial intelligence at the edge, I.E, the decentralization of artificial intelligence due to the size, and power of chip technology, meaning that we have the processing power available locally to deliver more timely analysis of the data for very time sensitive health situations.

There are also some important regulatory trends to consider. On January 1st, 2018, the centers for Medicare, and Medicaid services, CMS, allowed providers to bill for examining patient-generated health data, such as that produced by medical IoT devices. This ruling will allow a market to form, involving device, and software companies, marketing solutions to providers to automate the way they view patient-generated health data, and build more anticipatory, and personalized care plans.

Software as a medical device is also an interesting consideration. The Food, and Drug Administration recently piloted a precertification program to accelerate the acceptance of digital therapeutics within healthcare. The FDA is attempting to address the gap between lifetimes in software, and consumer electronics typically measured in months, with the time span required to run clinical trials to demonstrate clinical evidence of benefit.

We still need these studies, we still need the evidence of benefit, but we need to do it in a more timely fashion. There are some interesting trends in the way that physicians are licensed. This trend is perhaps more impactful to telemedicine, but it still applies to the related field of continuous care. The Veterans Administration in some states allow providers licensed in other states to practice medicine.

The Veterans in the E-health, and Telemedicine Support Act of 2017, The Vets Act, allows a licensed health care professional of the Veterans Administration to practice his, or her profession using telemedicine at any location in any state, regardless of where the professional, or patient is located.

The regulatory shift in the FDA, Veteran's Administration, and state's laws related to telemedicine, and remote care are paving the way for more innovation. Recently, Kaiser Permanente disclosed that about half of all visits in its system were via some form of online interaction. Let's also examine some key information technology trends.

Artificial intelligence, specifically, deep learning, will allow for a much more effective means to make sense of the volume, velocity, and variety of the data produced by these our IoT devices. Deep learning algorithms can be built to make sense of the patient generated health data, and provide insights to patients, providers, and payers on what is working in continuous care, and what is not.

Connectivity is also very important. As 5G wireless is deployed across the US, and in other countries, we will have ample networking capacity to move data from the person's medical devices, to data centers for analytics. Secure data streaming is also important.

With the additional capacity suggested by 5G, we will see the rise of better security, techniques, unlocking telemedicine through smartphones, secure health related communication from devices, to the smartphone, and then ultimately to the data center, and secure chat, and alerting functions from the provider, to the patient related to healthcare advice, and recommendations.

The confluence of these trends from all four categories then suggests it's time for a move to continuous, anticipatory, and personalized healthcare. Current healthcare data, for example, electronic medical records, labs, claims prescriptions, are just the tip of the iceberg compared to the total set of healthcare data available now, and in the near future.

Some estimates have this traditional motive data to be only around 10 % of the eventual total. Genomics data, specifically, sequenced data, annotations, et cetera, represent perhaps 30 % of the eventual total. The vast majority, around 60 % of health data will be sensor data generated by health-related IoT devices, and other exogenous data, for example, demographics, socioeconomic data, environmental data, nutritional data, et cetera.

The extent to which we can integrate these healthcare data modalities is the extent to which we can better understand each consumer's health. Perhaps the biggest difference between the traditional healthcare system, and the continuous care model is the transition in the way we get healthcare data.

Currently, a person interacts with a medical system fairly infrequently. An annual trip to the family doctor perhaps, or maybe an unfortunate visit to the emergency department results in a burst of electronic medical record data entries, and perhaps some new lab results. The infrequent nature of healthcare data capture suggested above the timeline in the diagram results in very episodic, and reactive healthcare delivery.

Contrast that with continuous care. Look below the timeline. Therein depicted is a set of data entries from a person's medical devices such as the continuous blood glucose monitor that samples health data potentially every five minutes. There are many examples of IoT devices delivering healthcare data at a similar continuous streaming rate. This fundamental change in volume, velocity, and variety of healthcare data generated by medical IoT devices suggests a fundamental shift in the way we can deliver healthcare.

The data, plus the analytics, and other systems that we can put in place can achieve a fundamentally new mechanism for engaging a patient with his, or her healthcare, and developing a much more anticipatory, proactive, and personalized healthcare system. Although data is key to continuous care, device data alone is insufficient.

To get an accurate picture of the patient, the patient-generated health data from the devices needs to be integrated with other healthcare related data. This complete picture of the patient needs to be integrated into the system, specifically, integrated into the electronic medical record.

IoT alone is not going to disrupt healthcare, rather, IoT is going to generate the data that will catalyze an enhancement to the existing healthcare system. What is suggested by this view is that the continuous care model of healthcare relies strongly on the notion of the digital mean, or complete health record of the individual.

Even patient-generated health data from these devices integrated with the other health care data modalities, and integrated into the electronic medical record is not sufficient. However, the volume variety, and velocity of healthcare data generated by personal medical devices, when integrated with these other healthcare data modalities provides an excellent basis for training, and running artificial intelligence infused analytics that can discern patterns from this data.

These patterns can then be the basis for healthcare decisions for, and by providers in conjunction with patients, and caregivers. Without this complete picture of the individual, it is much harder to deliver effective analytics to deliver better healthcare. The continuous processing of the data results in a continuous stream of healthcare insights that can be communicated to providers, patients, and caregivers.

These insights can be brought to the physician by a SMART on FHIR integration through the electronic medical record. These insights can be summarized into a portal, or a heat map for the provider to continuously understand which of their patients they should pay attention to right now, which of their patients could most benefit from a timely outreach from the healthcare system.

These results can be communicated to the patient, or caregiver, or authorized family member by the provider using a combination of alerts, chat bots, virtual assistants, telemedicine contacts, et cetera. Of course, for some of these results, communication might go directly to the patient, or caregiver as configured, and authorized by the provider.

The kinds of interactions between providers, and patients becomes more frequent, more proactive, resulting in a much higher degree of personal engagement with one's healthcare, just like we saw in the story about Kirsten earlier in this presentation. Continuous care is based upon a continuous stream of patient-generated health data produced by healthcare IoT devices, and integrated with other healthcare data.

This more timely, and complete picture of an individual enables the creation of patient insights from for example, the next generation of artificial intelligence analytics. Continuous care represents a shift in healthcare towards precision medicine. From reactive care based upon sparse, point-in-time health data, to continuous, anticipatory, and personalized care.

Continuous care enables a completely new level of patient engagement with their healthcare. Kirsten's story that we saw earlier in this presentation is possible to achieve, but there's a lot of work to be done across our industry to realize concrete solutions to problems like Kirsten's. Here are some considerations that need to be part of an overall IoT, and patient-generated health data solution.

We need infrastructure to cope with the volume, variety, and velocity of streaming IoT data in a secure fashion. We need an integration of this stream patient-generated health data with other healthcare data modalities. We need integration with the electronic medical record, we need the analytics to generate timely patient insights that act as the basis for continuous, and meaningful engagement of the individual with their healthcare choices, and decisions.

We need integration of the data across device, and treatment, and silos. We need visual to help providers treat individual patients in patient populations. We need secure communication channels between providers, and patients to enable proactive, and timely health interventions.

To me, the exciting thing about IoT, specifically patient-generated health data, is it's ability to transform healthcare. Think about what we could do if we had a continuous, and nearly real time telemetry about an individual patient's health journey. Instead of generating the data only when a patient is seen by a clinician, we could get updates every day, or maybe every hour, or perhaps every five minutes depending upon the device, and the time sensitivity of the data.

We can examine trends more closely. We can examine fine grain measurements that suggest the onset of, or worsening of a condition, and forearmed with this warning, we can take actions to prevent, or forestall the course of a patient's condition. The volume, variety, and velocity of patient-generated health data provides the opportunity to shift from episodic, to continuous care, from reactive, to anticipatory care, from anonymous, to personalized care.

Imagine what we could do if we had the technical infrastructure to make this happen. Imagine the new approaches to engaging patients in a more meaningful way to make better, healthier choices. Imagine how we could use patient-generated health data to fundamentally improve chronic care management, imagine the new approaches to remote patient monitoring.

Hopefully, today's session has inspired you to think about healthcare in a new way. Be sure to check out Optum.com/iq to learn more about how data, analytics, and technology are coming together to change healthcare for the better, and thank you for tuning in.