A couple of weeks ago, I was invited to speak on a panel at the Brookings Institution discussing the 5th-Generation (5G) wireless revolution, how it will enable the Internet of Things (IoT) and the implications for healthcare. In the paper that spawned the Brookings panel, Darrell West noted that 5G networks will incorporate cloud storage and a distributed computing model into a true Internet of Things, where billions of devices will share data in new ways.
The possibilities for identifying important health trends and intervening at just the right time to affect behavior — using everyday objects and systems — opens the door to all kinds of possibilities for improved health. But for me, that idea is positively frightening, given the current state of interoperability in healthcare.
The IoT only works if there are standards. An essential function of IoT standards is to allow devices to identify their capabilities to each other, and provide basic information regarding format. Several organizations are working on this, but none seem well-versed in healthcare needs.
As low-cost, consumer-grade sensors are embedded into devices without regard for the unique needs and requirements for healthcare data exchange, will we experience trouble? Will the 5G healthcare IoT become a ‘race to the bottom,’ where data of dubious or unknown quality will power our new healthcare ecosystem?
Take, for example, a connected weight scale. Even if a device collects data accurately, the way the scale transmits data can enhance or nullify its accuracy. For general wellness, this probably doesn’t matter. But for monitoring a condition like heart failure, it does.
One connected scale that sells for under $100 and published accuracy and precision specs said this: “Body weight: ±1.1lbs (11lbs~88lbs); ±(1%+0.2lbs) (88~330lbs).” This 1% error is a potential issue for monitoring heart failure, where a half pound weight gain can be a marker of significance. But what happens when that scale reports an individual’s weight to a larger data system? With today’s proprietary APIs, most companies report a weight number without any measure of accuracy or precision, and in my experience don’t even adhere to the convention that a reported number is accurate to + or – its least significant figure.
Here’s an actual example from one manufacturer’s API guide, showing numbers of two different precision levels in one sample response:
For example, the number 73 could be interpreted to mean anything between 72 and 74. But 72.5 implies would mean any number between 72.4 and 72.6. And by the way, what units are these, pounds or kilograms?
Precision matters in medical measurements; what happens when we find weight sensors embedded in beds, chairs and floor mats?
These are the kinds of problems that standards solve. Although the structuring of data transmission may seem complicated, the good news is that we have spent decades figuring out the best practices to accurately communicate data, based on trial and error.
So, how do standards solve the weight problem noted above? Using IEEE 11073 as an example, by specifying that the measurement is recorded as a 16-bit number with an implied precision and using defined values for transmitting the scale’s resolution, it is possible to reconstruct any measurement with no loss of precision.
The Continua Design Guidelines (CDGs) ensure that medical-grade data can be collected and transmitted whenever and wherever needed. The foundation of the CDGs are the FDA-recognized IEEE 11073 personal health device standards. Within this series, the communication of data precision is mandatory and embedded within the datagram format.
If you make devices using Bluetooth, as most modern devices do, implementing the correct standards is easy: start by reading the Personal Health Devices Transcoding whitepaper available from the Bluetooth SIG. If you buy personal health devices, ask the manufacturer if they comply with recognized standards, look for the Continua Certified logo or find the device in our Certified Products Showcase, your assurance that the device can provide medical-grade data.
Together we can build a truly medical-grade Internet of Things.
Robert Havesey, MS is a vice president with the Personal Connected Health Alliance