by Nahum Gershon
The Internet of Things is already affecting the choices we make on a daily basis, including those related to our health. Yet challenges still remain in both the design and manufacturing of devices in order to improve accuracy and effectiveness. LSTC member Nahum Gershon weighs in.
When I used to leave my home, I had to manually turn on the security camera and dim the lights. When I returned, I needed to manually turn the camera off and increase the intensity of the lights. Following the advice of my friend, Kyu Chong, I have now managed to automate this process by letting the home network router send a message to the camera and other devices that my cell phone (and thus me) is leaving or entering its range. Such interactions among wearables (the cell phone could be considered a wearable, not just a command and control center!), humans, and our Internet of Things (IoT) devices have the potential to transform the plethora of all of these devices and entities into a single ecosystem: Wearables (including implantables), Humans, and Things—The Internet of Everything!
Heterogeneous Components But One (Potentially Complex) Eco-System
It is important to view these entities as part of one system because these components—wearables, things, and humans—all could function together as a system, interact, and relate to one another in ways that were not possible before. How would you describe, for example, an interaction or rather, an interplay between a human being and an artificial organ embedded under his or her skin? These relationships could be quite different, physically and psychologically, from the interactions between a person sitting in front of his/her computer back in the 1990s. We thus need to develop new methods to describe these relationships and not use the same old metaphors (for example, “human computer interaction”) that could cloud our thinking as we move forward.
Moreover, many IoT and wearable devices are connected and frequently controlled by specific hubs. These hubs function as the command and control centers of their specific devices. This variability creates a heterogeneous system consisting of groups of IoT devices and wearables, each controlled by a specific hub (with potentially separate apps, rules, and automation schedules), and humans. For these composite systems to function properly and safely, appropriate and effective strategies need to be developed so that the different components will work in harmony. One possibility is another consolidating layer with an integrative app on top of the plethora of the heterogeneous components.
In addition to the challenges on the system level (for our purposes, a system is a “community” of entities), there are challenges at the single component level, including devices, hubs, wearables, and humans. Following are some examples.
The Internet of Things (IoT) and wearables are very much in vogue these days. New devices and possibilities are released every day. For example, we have a whole spectrum of fitness trackers, many of which are worn on the arm, but there are also some with belt clips following the core motion of the user (core trackers). The trackers typically provide numbers of steps, flights of stairs, calories burnt, etc. We humans, in turn, usually tend to accept numbers as accurate. But, how accurate are these numbers? To find out, I carried out some basic experiments and the results are noted in Tables 1 and 2.
Table 1. Different trackers give different results
(for 100 steps)
In this first experiment, results differ up to about 10% from the real value (100 steps). In addition, core trackers gave more accurate counts than trackers positioned on the arm (possibly due to contributions made by arm movements).
Appropriate Measures: Flights of Stairs or Elevation?
A second experiment involved two identical core trackers that measure elevation change (instead of flights of stairs climbed), which were placed in one pocket. I walked half a mile on a flat surface. The results can be seen below.
Table 2. Feet climbed measured simultaneously by two identical core trackers on a flat track
|G Tracker 1||G Tracker 2|
|Feet climbed at 0 miles||324||451|
|Feet climbed at .5 miles||339||553|
|Difference in Feet||15||102|
The reported changes of elevation while walking on a flat surface and the difference in elevation reported by the two identical trackers are rather stunning and inaccurate!
Most other trackers measure cardio effort by tracking flights of stairs climbed (in a short period of time) and not a general change in elevation throughout the day. Which one is better for this purpose? If we take the immediate time it takes to change elevation out of the equation, we are confronted with the following estimate:
If I walk 10,000 steps a day, this is equivalent to about 4.46 miles. If my aim is to also climb 20 flights of stairs each day, and if each flight of stairs is on the average 8 feet, this is equivalent to climbing 160 feet a day or 36 feet per mile on average.
This is quite a low slope and if you would climb it gradually, it is unlikely to strain your heart! So, having a measure of feet climbed without constraining the climb to be done in a short period of time might be practically meaningless as far as cardio fitness is concerned.
Responsibilities of Designers and Manufacturers
The list of issues with consumer-level wearables and IoT devices can go on and on. For example, it would be helpful to have an on/off switch so the tracker will not acquire false physical exercise, such as data while on bumpy rides. In addition, paying attention to physical comfort considerations for wearables and providing instant response (and thus the expected instant gratification) to an activity could improve the experience for a user. To have this eco-system function properly, designers of fitness devices, wearables, and hubs need to be both knowledgeable and considerate of human capabilities and desires, as well as the pitfalls of current designs.
Consumer-level wearables and most IoT devices are not medical devices so strict standards are not imposed. However, consumers tend to sometimes take seriously the information provided by these devices, especially if they are related to health. Thus, it should be expected that tracker, device, wearables, and hub manufacturers will test their devices under different conditions and publish the results so that people would know what to expect. These tests should include accuracy, reliability, and operations under various environments (different hubs, routers, etc.). Tests should be carried out not just at the beginning of development, but also for each software version release to prevent incidences similar to a case where a “smart” thermostat stopped working after a recent software update, thus leaving many people without heating in the winter time .
Developing full-fledged standards for consumer-level products such as things/devices, wearables, and hubs might not always be required or practically feasible. Even so, perhaps some guidelines could be developed for measuring accuracy and reliability, and functioning of devices in a homogeneous environment as well as in heterogeneous systems. Such guidelines would be best developed in mixed groups of engineers, life scientists, health professionals, and designers. The same holds true for working on full-fledged standards for medical devices, wearables (including implantables), and hubs.
In the near future, the IEEE life Sciences Technical Community plans to foster opportunities for these mixed groups of professionals to come together and work on solving these complex issues. Input from the life science community will be crucial in developing effective solutions to the challenges the IoT represents.
The first IEEE International Conference on Internet-of-Things Design and Implementation (IoTDI 2016) will be held in Berlin, Germany from 4-6 April 2016, and the IEEE World Forum on Internet of Things will be held 12-14 December 2016 in Reston, VA. You can also find more general information on the Internet of Things at iot.ieee.org.