A good friend of mine is a passionate runner, even more than me being a passionate cycler.

He has run most of the famous marathons in Europe, including the 100km ultramarathon of Bienne a number of times.  He has been a diabetic for a few years now, and as a Swiss technophil, he has tried out various gadgets to monitor his blood sugar, and to keep on running without risking his health.

 

I still remember the times when you had to prick your finger, capture a drop of blood, and have it analyzed in a small handheld. I was surprised to hear that this method, though inconvenien,  continues to be more reliable than some modern sensors that measure directly.

 

My friend still uses this method to calibrate his more modern IoT gadgets. He showed me a small plaster that covers a sensor that continuously measures blood sugar, with the big advantage that it can also show trends.

 

He explained that for a longer run 5.8 as a single measuring value does not tell him a lot. If his blood sugar level is trending up, he is safe and can start off. If it is trending down, he should rather take a bit of dextrose to make sure he will not faint while running.

 

Boundaries of reliability

I asked him why he does not use the little machine together with the insulin pump to directly control the level of blood sugar without him having to eat dextrose etc.

The machine is already able to do this.

 

The tricky bit, I learned, was the responsiveness. Most sensors work perfectly precise as long as you are not engaging in heavy exercise. For a runner, it is essential that the sugar measurements do not lag behind reality – or you may reach critically low sugar levels and faint.

Some machines allow users to enter additional information like before or after exercise. If you plan to run an ultramarathon, this is a little too blurry.

 

So, my friend will not blindly rely on the machine reading, and complement the information with a human’s broader view – and either slow down a bit or take a dose of dextrose . Connecting the insulin pump and creating a closed loop system feels still too risky for unusual circumstances like an ultramarathon. (And as a last resort, I know where his glucagon injection is located.)

 

Would you be able to relax and read a book riding in an autonomous car on a Geman autobahn?

Maybe five years from now? I am well aware of my own bias that being in control behind the steering wheel would feel safer.

The Dunning-Kruger-effect, and many subsequent studies, nicely show that the overwhelming majority consider themselves better than average drivers. Thinking of German autobahns I can absolutely confirm this from my experience.

So, at what point in time will a machine drive better than a human? Already today?

 

The human plus machine advantage

Many industrial IoT proof of concepts and pilots start small, and I firmly believe that it is right to do so. On the other hand, some answers rely on the bigger picture, or require at least a few additional pieces of information.

Consider a vibration sensor on a crusher, or on a rolling mill.

The vibration pattern by itself gives a partial picture. I remember similar heated discussions around benchmarking OEE across machines or plants or teams. What makes performance comparable? Is speed, quality and availability alone telling?

The hardness and quality of the material crushed, or the grade and target surface quality of flat  steel add to the picture. Add the customer of the steel plate and the expected profit and you get an even complete view of the world. Blend in cost of complaints etc.

 

None of this requires human experience – but a good data scientist and an experienced operator can adjust the boundaries and scope of a model, and will, in my opinion, achieve much better results.

 

Will IoT get me the Tour de France’s yellow jersey?

I seriously considered adding torque sensors and heart rate to my bike computer, complemented with height information and ideal pedaling frequency per steepness of ascent.

Unfortunately, the muscles pushing my bike forward are still mine. There’s a lot more than IoT keeping me from winning the Tour de France. But I will enjoy the ride anyway.

To report this post you need to login first.

Be the first to leave a comment

You must be Logged on to comment or reply to a post.

Leave a Reply