Predictive maintenance is one of the oldest and most tested uses cases for the Internet of Things (IoT). For years, we’ve been able to analyze incoming data from sensors embedded in machines and make decisions about whether or not maintenance activities should be executed.
Typical scenarios have focused on things like wind farms, oil rigs, and fleets of trains. And while there’s plenty of excitement and new developments in these areas, what’s grabbing a lot of attention today is how predictive maintenance can be applied to new scenarios.
For example, in an earlier blog, I talked about predictive maintenance for autonomous vehicles – how sensors can send out data on the status of parts and components, allowing manufacturers to analyze this data to predict part failure and, thus, avoid breakdowns.
Yet, even this scenario keeps us in the realm of machines – because, as sophisticated as it may be, an autonomous vehicle is still a machine. But what if we could take the same general idea of predictive maintenance for machines and apply it to our bodies? Call it preventative maintenance for people – or just predictive healthcare. The reality is that in many ways, we’re already there.
Understanding in context
One of the advantages of predictive maintenance for machines is that incoming data about what’s going on in the moment can be analyzed in the context of historical data about the same machine. Let’s say an HVAC machine on the top of a hotel in Seville – where I live – sends out a high-temperature alert.
In and of itself, this may be cause for concern. But when you realize that the machine sends out the same alert every month at the same time – well, maybe it’s not so concerning. Maybe the HVAC unit runs continuously for eight hours on the first Monday of every month to help cool a large conference room for the packed monthly meeting of the Seville Dog Walker’s Association.
Or maybe there’s another reason. The point is that in such a scenario, the high temperature alert is understandable and predictable in context – and thus of little concern. It would be nice if we had something similar for healthcare.
More than a snapshot
On a typical trip to the doctor, you sit in the waiting room for 10-15 minutes with other people, some of whom are likely sick. When you finally see the doctor, you’re thinking of the next appointment you have across town in 30 minutes, so your anxiety levels go up.
My sister Mary was recently diagnosed with high blood pressure because her blood pressure measured 150/89 in the doctor’s clinic. The doctor advised her to get a connected blood pressure cuff and take regular measurements. When she did, it turned out her blood pressure was 108/75 – completely normal. She was suffering from what doctors call White Coat Syndrome.
Her high blood pressure reading was understandable in the context of her being in a doctor’s office, much like the HVAC data was skewed by a temporary situation. Wouldn’t it be great if the doctor had more than a snapshot of (often misleading) blood pressure data to work from? Wouldn’t a whole bunch of relevant historical data be better?
With the smartwatch on my wrist, I can now share three years of data with my doctor. Now she can see things in context and treat me more effectively. I think it’s only a matter time before her office can take my sensor data in automatically, over the cloud. This will make my yearly checkup more productive. Instead of figuring out what the problem is (if there is one, hopefully not), we’ll be able to focus on what to do about it.
A business network for health
As with so many things IoT, this is only the beginning. But let’s step back for a moment and look at the notion of an asset intelligence network (AIN).
Think of it as a business network application. All of the data (metadata, specifications, bills of materials, etc.) that goes into the creation of a device (compressor, coffee machine, car, etc.) can be stored in a central location.
When connected to the asset intelligence network, the device can push out real-time data that describes its state at any given moment. When the device owner allows access to this data, the manufacturer can then analyze it in conjunction with other data from other devices – making product improvements that can then be pushed out by way of the same asset intelligence network.
In fact, nothing is stopping device owners from sharing their data with anyone they wish – such as a service vendor or an insurance company. If a device goes out of tolerance for some reason, the service vendor could receive a notification and schedule an appointment to service the device automatically. Or an insurance company could set rates according to actual device usage data.
Returning to the theme of health, what if we took this idea of an asset intelligence network and applied it to our own bodies? What if we had a “people’s intelligence network” where a device like my smartwatch publishes my health data into a trusted cloud application? When my device senses high blood sugar, for example, this data is analyzed, not only in the context of the unique moment mixed with my own personal health history, but also in the context of similar data from potentially millions of people.
Based on this much larger dataset, the network could then contact my service vendor – in this case, my doctor – and make an appointment if necessary. Yes, this would be convenient. But more importantly, it would move us away from making medical decisions based on poor data and the intuition of physicians, toward something often heralded but seldom achieved – real evidence based medicine.