Data is the raw material of artificial intelligence, meaning that it will be increasingly critical that the sensors collecting that data are reliable and accurate.
Nowhere is this more true than in eHealth, where inaccurate data is useless at best, and deadly at worst. This is why there is still a big market for extremely expensive medical monitoring equipment, but I see signs everywhere that this is starting to come to an end.
This also explains the problems that the likes of Fitbit, Apple, Xiaomi and Garmin are having, as the data they generate is of such low quality that it can really only be used for recreational fitness.
I see two ways in which the data that these sensors generate can be improved.
1. Improve the quality of the sensors themselves
If an optical heart rate sensor can gather data as reliably and as accurately as an ECG, then this would have substantial ramifications for cardiac medicine. Not only could the equipment costs be slashed, but high-risk patients could be continuously and non-invasively monitored, allowing many cardiac events to be predicted and stopped before they occur. The sensor industry is feverishly working on this with the latest launches promising more and more accuracy and detail.
Despite this, I have yet to meet a cardiac sensor company that is claiming that it can hit the kind of quality that would allow it to be certified with the FDA.
The same is not true in blood pressure where small start-up Leman Micro Devices is making some bold claims. It has come up with a tiny blood pressure sensor that can fit onto a smartphone which it thinks is very close to being good enough to measure blood pressure at a medical grade with FDA approval.
2. Create intelligent software that improves the quality of the data
There are many examples of algorithms being used to meaningful conclusions from low quality data both in and out of the medical field.
In automotive, retro-fitted vibrations sensors are being used to track the condition of tires, wheels, shock absorbers, brakes and the steering wheel (see here and here). This is not because a great sensor has been invented, but because these companies have worked out how to interpret data that most people consider to be random noise.
Phillips is also quite good at this, which is why its health watch is recognized to be generating good quality data, even if it does struggle in other areas.
Hence, I see the road to accurate data being paved by eroding the problem from both ends, combining better hardware sensors and much better software to interpret the signals.
This is crucial as sensors are the eyes and ears of the machines upon which the world increasingly depends. Consequently, I think that sensors will remain an area of intense investment and an area where I would want to be invested.
The issue of course is to separate the solutions that have real prospects from those that are merely riding the wave of hype and easy investment.
This article was originally published on RadioFreeMobile