Which Grasses Aggravate My Allergies?

It seems like I’ll be hitting around 270-280 sneezes this year from where I’m at right now because after the grass pollen season has ended, I rarely sneeze during the fall and I’m around 260 now. So, you can see there’s not much data to work from anymore.

So, I ventured into tracking a precursor to sneezing and that’s itching in my nose, and I’ve actually been doing that for some years. But this year I’ve been using some new instrumentation and I started using this in May, and I’ve been making 6,586 observations until yesterday. So that’s around 49 observations a day, so that’s pretty high frequency of tracking this subjectively of this perceived phenomenon.

But the distribution is something I won’t come into during this presentation, but I’ll just show you the instrumentation here. So this is the one button tracker and Jacob E. Larson, we are building these ourselves now because the commercially available, the hardware that we’ve been experimenting with didn’t really live up to the specs that we found that’s required for this kind of instrumentation. So, we’re doing a session later on, actually two sessions later on today about subjective experience tracking and then also how can you actually build these kinds of devices yourself.

So, what I do is every time I feel an itching nose or tingling in my nose, I press this button. And then on a drive in here, inside the device it will be stored in a CSV file with a very accurate timestamp down to milliseconds because we put a GPS chip in here, so we get very accurate time from the GPS but not location as such.

Then during the summer where I’ve been using this instrument all day, I’ve also been running in the park. And I started noticing that it was as if when I was passing certain locations in the park, I would be making more observations.

So that’s actually this experiment that I’ll be talking about now, which is I then tracked itching in my nose alongside high-resolution positions and then also I did a couple of rounds time-lapse photos of the vegetation that I was passing while I was running the same loop in the park.

And I used a GoPro camera for taking photos of the different locations I was passing. And then I’m also creating a new kind of running app, and because I’ve created it myself I know that I can get locations, the positions every five to eight meters. So, it’s very high-resolution positions. So, I had these different streams of time series data that I’ve been looking to merging.

So, last year in Amsterdam I did a talk that was called Over-Instrumented Running, so I guess this one is even more over-instrument running than it was last year because it went somewhat more crazy.

So, I’ll show you here, I made some heatmaps by merging these streams of data together. So, these are two workouts where I also did, I also wore the GoPro camera. And you can see the first workout I did 139 observations while I was running. So, I was carrying it in my pocket here and I would just press it immediately when I felt it, so it wasn’t any hassle.

I’ve been doing more than 120,000 observations over the years using a Smartphone app, and that wouldn’t be a way of doing this kind of tracking because it would interfere with the flow of the running experience and it might even be a safety concern right to pull out a Smartphone when you’re running.

So, this is how I created to do it, and then you see the second workout. There were only 79 observations and there are like four hotspots emerging on the heatmaps. And those hotspots that I can see the data actually corresponded to what I was noticing myself while I was out running. So it seems my own experience was very well calibrated with what the data showed afterwards.

Then I’ll just show you something very interesting here which is these are the official grass pollen levels in Copenhagen at that time. And it turns out that on the day when I made the most observations the Danish level was 26, and the day where I made the least observations the level was 122. And this is actually something I saw already, like 10 years ago when I started doing allergy tracking was that my own experience of the allergies was kind of un-correlates to the official pollen count. So that was some of the things that got me started like looking into.

So, I’ll show you some vegetation at these different hotspots. This is the vegetation at hotspot number one, and imagine this park, there’s a big lawn in the middle which is grass, which is what I’m supposed to be the most allergic to. But this is the vegetation hotspot number one, so you can see like bushes and like wild grasses and stuff there, kind of a more kind of closed spot. This is hotspot number one again, and the same kind of vegetation.

This is hotspot number two, and you see to the left this kind of vegetation is coming up again. This is hotspot number three, well, it kind of looks the same, kind of closed. We have some bushes, we have some wild grasses going on. And this is hotspot number four is kind of like the same.

And then I asked my parents for some really old data that we don’t have much of because this next thing I’ll show you is from the early 80s. this is my sister and I in the garden of the house that we moved to in 1981, and I got my allergies the year after.

So, it seems like my immune system is kind of clinging onto the lesson it learned back in 1981, even though I’m not feeling my symptoms as I had been for 30 years. But it seems that learning, it’s tagged something that was present there. It’s still remembering that learning, and so my work has been trying to un-learn and teach my immune system not to hold on to some associations from back then.

Then a learning too is that even though I have been part of building the two tools that I’ve been using, and I’ve been in control of the data performance, especially the time performance, merging observation streams is just not brilliant. I had to do a lot of coding to merge those streams of data to present those things, those heatmaps I showed you today.

So, it says something about the maturity of the tools that we have and exposed to. I have to use my skills as a software engineer already like 10 years into kind of make these kind of analysis. I think that’s kind of something to ponder about, and we’ll talk about that later today when we talk about the open stack.

Then this is also for me evidence of kind of one of the ideas I’ve been pursuing for the last 10 years, which is the kind of biometrical prominent explanation of allergies that is the misfiring of the immune system. I don’t think that’s correct.

I think my immune system made some very specific inferences back in 1981 that was based on kind of the bound of rationality and the data that it had at its disposal, and it’s still kind of clinging on to that difference. But I don’t think this was a misfiring. This was just, there were specific circumstances that made my body make some decisions that it’s still clinging onto.

So, I think that I’ve learned that my immune system is a fast learner but a really slow forgetter. So thank you guys.