From hydration to ovulation, health trackers keep tabs on nearly 1 in 4 Americans. But wearable devices like these are just one piece of the ever-expanding medical "internet of things" - a universe of internet-enabled devices, applications, wearables, and more that collect, share, and analyze our data.
UC San Francisco Associate Professor of Medicine Sandeep Kishore , MD, PhD, MSc, is part of a joint UCSF and UC Berkeley team preparing to pilot wearable devices to help treat some people with diabetes and high blood pressure at UCSF. We asked him about wearable health technology and what the future holds for these ubiquitous gadgets.
How is wearable technology already intersecting with our lives?
Many people are likely to think about fitness trackers that count your steps, monitor your heart rate, and even take your blood oxygen level. But there are also smartphones. You can think about a pharmacy or a message from your health care team on MyChart. That's all a form of mobile health.
One of my favorite examples is sensors, think about blood pressure cuffs or continuous glucose monitors that connect to your phone and provide the data instantly. If you're unfamiliar with glucose monitors, these are little patches with a very tiny needle that usually go on the upper arm and can sample your blood sugar levels about every five minutes.
We are working here at UCSF to find new ways that the information can be relayed to your health care team and be made actionable.
How could wearable technology fill a health care data gap?
As physicians, we often only get a snapshot of what's happening for a patient, like when we take your blood pressure at a clinic visit. It's surprisingly difficult to get accurate blood pressure readings in a clinic.
Patients may have just had their morning coffee and taken the stairs, or they might be a little anxious - all of these can temporarily increase your blood pressure reading at the doctor's office. It's what we call, "white coat hypertension."
Wearables, like electronic blood pressure cuffs, could record your blood pressure 365 days a year, sending the data to a secure system that could give your physician a real window into your blood pressure over time, not just at six-month check-ups.
Where do you see the future of wearable health technology in five years?
- Wearable tech has to get easier for patients: Finding ways to, for instance, use a smartphone one day to take your blood pressure through your camera phone or, in the future, check your heart rate via a Zoom video recording, capture my attention because they don't take much effort.
- Digital twins: We might one day be able to build a "digital twin" of patients - a computer model of their health that gives their care team an additional tool. It's still early, and I haven't yet seen it clinically validated, but I'm intrigued by that in the next five years,
- It'll be the bouquet, not the flowers: It's an idea I learned from UCSF Professor of Medicine Ida Sim , PhD, MD. You can imagine a number of gadgets focused on just one condition. This is going to lead to data overload, and, as a physician, more data than I'm going to know what to do with. Pulling these data streams - the flowers - together into a bouquet, to make it useful, simple, and scalable is going to be the secret sauce.
What are the challenges to all this data?
The sheer volume is huge. Each patient can generate gigabytes of data per month, which is a data processing challenge.
The other issue is with aggregation and standards. Different devices track data differently. Companies have proprietary algorithms behind the data, which they sometimes lock, so harmonizing and combining data to make sense of it will be a challenge.
Could artificial intelligence deal with this deluge of data?
Yes. Artificial intelligence has the potential to sift through the firehose of data to detect new patterns in diseases. Those patterns could help us understand what's behind symptoms or even what's driving disease. It may also help us predict the risk of certain conditions. The goal would be to turn all that data into clinically actionable alerts and interventions. At UCSF, we're working to find ways data from wearable technologies can be relayed to your health care team to help them support you and, together, make better decisions about your health.
What would this look like in real life?
I think about a patient I had on the wards recently. She was in her 30s and had type 1 diabetes, meaning that she requires frequent insulin to manage her blood sugar. Unfortunately, she ran out of insulin and presented to the hospital nearly comatose.
In many ways, she was hidden. Her roommate was the one who found her slouched in her room. If my patient had some sort of passive blood sugar monitoring, we could envision a day when that data could be part of a feedback loop between her and her health care team. Imagine if it sent an alert to a physician or a pharmacist monitoring a dashboard? Or maybe initiated a call or text to her phone that - if she didn't respond - would trigger an emergency response? Maybe, we could have prevented this from happening.
Will AI replace doctors?
No. Clinical insight is still very essential. It's not the case, in my mind, that a data scientist or an AI expert alone can take a bunch of data and generate a clinical insight without any clinical experience. To build tools, you're going to need cross-functional teams with developers, clinicians, patients, UX, designers, etc.
How is UCSF charting the future of big data as it relates to health care?
UCSF and UC Berkeley are working together to bring wearables into the clinic. We've partnered to build an open-source platform called JupyterHealth to bring health data and AI together for diabetes and high blood pressure, some of the most common conditions. The platform uses AI models to surface key insights for clinicians and patients in near real time to help manage these conditions better. Our goal is to leverage this unprecedented level of data to help clinicians and patients make decisions now that otherwise might have taken months to years with typical monitoring.
How does UCSF ensure that the AI solutions it designs and studies are safe, secure, and ethical?
UCSF has a rigorous system of checks and balances that starts long before any study does. As a physician-scientist researching AI, I have first-hand experience with this process. We have a new Health AI Oversight Committee of experts that reviews the projects to ensure the AI we produce and study is trustworthy, and that it's secure, fair, and protects people's privacy.
Researcher must submit detailed research plans to our institutional review board. This expert committee must sign off on any research that might impact or involve human participants to ensure that research is conducted safely and ethically.
