Taking the measure of wearables

All wearable devices measure signals, but do they do it accurately? We spoke to metrologist Dr Annette Koo to find out.

One thing that all wearable devices have in common is that they detect and measure signals from the human body or its surroundings. With various studies showing that these devices are not always as accurate as you might expect, it’s clear that they have a measurement problem.

Measurement is at the heart of almost everything we do in our daily lives – the scales we use to weigh our food rely on an accurate definition of the kilogram, the power sockets that charge our gadgets wouldn’t work without measuring electrical current, and we’d quickly find ourselves lost without the time signals that power GPS satellites.

The same is true for those in industry, too – every high-tech device or product that’s come to market has been through a rigorous validation procedure, to ensure it complies with a set of performance and safety standards that are defined by measurement. In some sectors, the need for accurate measurement is critical – an aircraft manufacturer precision-machining components, an electrical engineer designing a new national grid, or a radiation therapist selecting the correct dose for their patient. Without reliable, globally-standardised measurements, these industries would fail.

Thankfully, there is a science to measurement. Called metrology, its role is to establish and realise the internationally accepted units of measurement (built on seven base units) and to link measurements back to reference standards. And the metrologists (measurement scientists) that support New Zealand’s labs, industries, businesses and technology providers, are all found in one place – the Measurement Standards Lab (MSL), in Lower Hutt.

So, to better understand the role that measurement can play in wearable devices, we spoke to a metrologist – Dr Annette Koo, Manager of the Temperature and Light teams at MSL.

What does a metrology lab do? 

We make sure that measurements made in NZ are the same as those made internationally, to support trade and research and development. That involves tracking down any sources of error that might be influencing our measurements, as well as providing direct calibration services to our clients.

We work with a range of sectors – from companies trying to demonstrate compliance with standards, to those carrying out research projects on photoluminescent lighting and horticultural fabrics for crops, all the way through to those using digital rendering techniques.

What would a wearables start-up have to think about in terms of validating their device?

First off, they should look at whether there are any documentary standards that they're going to have to comply with before their device comes to market. And, they should consider if there are specific measurement parameters that they'll need to have control over.

More broadly speaking, they want to be able to show, from the beginning, that the measurements that their sensor delivers are going to be comparable over time. So that the result its gives is not influenced by the temperature, aging, vibrations, or whatever factors might apply in the context of their device.

A good test of a device is whether the measurements compare with those taken with other devices that are already being used internationally. Can we compare the temperature that you are measuring with the temperature measured by a non-wearable reference device?

And it’s important to remember that sensors will always need to be recalibrated – they don’t stay accurate forever. So, when you’re designing a system, try to make sure that the fundamental sensor remains accessible for calibration. 

Why is it important to calibrate a device?

The key word for me is always trust. You want to trust your own claims, and you want your customers and investors to trust them too. If you’re working on medical devices, bad measurements can be a question of life-or-death. So, measurement is about de-risking your project, and helping people to make better, more informed decisions.

Here’s an example of when a lack of calibration caused a problem. We were approached by a water company who used UV lamps as part of their treatment cycle. Over time, their detector began to show that they weren’t producing enough UV light, that they were degrading, so they kept replacing their lamps at a huge cost. But the bulbs weren’t the problem. They hadn’t calibrated their detectors, so as a result, they wasted considerable time and significant resources.

Should we trust measurements made by wearables?

I think it's good to push technology to the limits, but we could better asses the uncertainty of measurements given by wearable devices. And by that, I mean a scientifically-determined analysis of how far you can trust a specific measurement.

So yes, push, but always provide an additional measure of how reliable that number is, so that people can take this extra information into account when they make decisions. It’s not about saying ‘this one is bad, this one is good’; it’s about giving people a tool with which they can use the device.

Uncertainty analysis is something that MSL are really interested to help with. And that really means helping manage user expectations by evaluating just what level of reliability it has. It is not about trying to lower the uncertainty so much as understanding it, and making it available.

When should a company come to you for advice?

We find that people usually only come to us when there is a problem, and often when something has been a problem for a long time. So, we would encourage people to put in quality checks from the beginning, so that they don't end up potentially wasting resources before they determine the source of the issue. 

If you're getting into in the business of wearables, you should start by talking to the Measurement Standards Laboratory. Download their Measurement Solutions for Wearables flyer to see how they can help.