On the shoulders of giants
Tomorrow I’m doing something for the first time, ever. I got an email yesterday with the good news and while we don’t have all the equipment we need for the project, we can still get some good data and an early look at what the data will look like. I’m so excited I’m shaking, I barely slept last night, and my mind is swimming with the possibilities. This is what it’s like doing research on the edge.
A while back I had an idea, a “big idea,” which excited hospital-PI to the point that the more we talked about it the more money he threw at me to make it happen (more). I’m happy to say that in my original post it cost nothing to do and the final design was cheap enough that I built two and could’ve paid out of pocket for it and not lose sleep over it, if I felt so inclined.
It’s been months, literally, since I built and tested my prototype device. And since then I’ve been making more equipment to help improve my odds of success, because there’s no substitute for noise free data (more). None of the equipment we’re using was made to do what we’re about to do with it, mostly because no one, that we have been able to find in the literature anyway, has done what we’re about to do.
The big difference between what I’ve been calling big idea and my “super secret technique” (SST) is that big idea is much lower risk than SST, both will be amazing if they work, but BI has a much higher chance of success and will give us a higher information rich dataset to work with. One that I will have to make sense of once we collect, so that’s no pressure… at all.
Yesterday I got the email letting me know we have our first participant for the project!
In preparation for the big day, I plan on going over some of the SST data like I mentioned yesterday. This time though I need to start writing a bit of code in anticipation for the dataset we will be collecting. I’m sure hospital-PI will want a fast turnaround with this and there is about a million other projects I’m working on right now so I will have to fit in a very complex, very time consuming, very hands on, data processing regimen into my already packed schedule.
A lot of the work will be simply because I have a very limited idea about what the data will look like. I already can anticipate different noise sources and I can come up with ways to filter that noise, but there’s a whole lot of hands on things that need to occur to make sense of it all. Basically I cannot just blindly let the algorithms do the work without first setting parameters for the process or, in the worst case, I could very well remove data that was useful. Doing it blindly, the best case would be the data are not cleaned and the noise would hinder any sort of conclusion I could be making from the data.
I also need to think about how to organize the data best. Luckily most of that can be done prior to collection and I can write code to sort it blindly for me without fear of any problems that may arise. That will save time later, but only a little. Unfortunately, there’s not a lot of prep that I can do to significantly save time. The only comfort is that as we collect more data, the faster the process will go… theoretically anyway.
If you’ve never done research, it isn’t totally uncommon for a almost complete reset on everything you’ve done only to go in a different direction. In fact, that happens so often I’m basically numb to the idea these days, it’s one reason “last paper” took so long (here).
So we’ve concluded that the equipment is mostly ready, the stuff I’ve made is (hopefully) high quality, and there isn’t a lot that can be done until data is collected. But… there’s more. We haven’t discussed the experimental design for the project yet. And there is (unfortunately) a good reason for that.
Being the first means covering all the bases. We don’t want to do one thing, we want to grab as much as we can because we’re greedy little data goblins and want it all. So I fully suspect that tomorrows experiment will run the full three hours (despite minimal setup) and we will collect all the data, everything we can from this. The problem is that the more we do, the less of each category of thing we can get (due to time constraints and the person being super exhausted after three hours of basically lying in a bed doing what we ask them to do.
That’s my main concern for tomorrow. I need to make sure, blindly, that we get a minimum amount of data that will be useable from each of the things we will be trying. That is easier said than done! I’m going to give it my best guess, because… and this is where things get even scarier, I’m the expert on what we’re about to do. Yeah, me, don’t ask me how the hell that happened, but here we are. Meaning I’m running the show and hospital-PI is watching and hoping I know what I’m doing (all the while I’m also hoping I know what I’m doing).
Still I try to remind myself that this is a high probability of success experiment. I’ve already pitched one (of multiple) paper topics for the dataset we will be collecting and even though tomorrows dataset will be limited due to not having the full equipment list we really wanted for this project, it will be incredibly valuable and when we do get the rest of the equipment this dataset will help guide what we do next, and more importantly how much we do.
Of course, this is a group effort. As Newton once wrote, “If I have seen further it is by standing on the shoulders of giants.” In fact, this project is as close to a perfect mix of the things I’m interested in and the things hospital-PI is interested in as we can get. It will be a radical shift from the things hospital-PI has done in the past and I’m very excited to finally get the chance to show off the weird things my brain comes up with. We’ll still need a lot more data before we publish, but this is huge for us.
Regardless of how it turns out, tomorrow will be a good day.