Dissertation data update 1

I’m being lazy so we’re just going to number my updates with regards to the dissertation data I collected, because I’m tired and I can’t honestly think of a better way to do it at the moment. This is subject to change if I come up with a more creative way. So a while back I collected a bunch of data, then this weekend I had my first chance to teach the people who assisted the first steps to processing it. It turns out there’s a long way to go with them.

Welcome, for those who don’t know me, a brief intro is once again in order. I’m a fifth year PhD candidate in neuroengineering. My BS and MS are in mechanical engineering, which I like to point out are totally different skill sets! Early this year I defended my dissertation proposal (here), so now we’re on the final countdown to graduation (hopefully). It’s been a rocky roughly four years, but the past year or so has really improved for me. I’ve started working full time in a hospital (here), I’ve won awards (here for example), got funding for my dissertation project (here), and DARPA found me (here). Not bad, not bad at all. But there’s still work to be done.

I collected a lot of data already, like a whole ton of it. If data had a weight to it, this would be heavy. But I wasn’t alone, I had some help (here). Since they were polite enough to help me collect the data, I wanted to return the favor by helping them learn how to process the data and make sense of it. One of them I know for certain needs to learn this so they can graduate one day. The other has different plans so they don’t need to learn it, but if they want to learn, who am I to stop them?

We had our first meeting this weekend and I suggested that after the meeting I would post some of the stuff I taught here. You’ll notice that the post date came and went without fanfare. Instead I did a post on how 9/11 changed us and what not. I’m not rehashing that post or even linking to it, because reasons. So if you’re following along I wouldn’t blame you for wondering if and when that post may come, especially if EEG data processing is interesting to you. Well it will come, but probably not for a bit. Because as I learned this weekend, there’s a bit of a gap.

Okay, a huge gap. They had no idea how to use our software of choice, MATLAB, at all. Instead of walking through the basics of data processing and loading the EEG files, I spent the bulk of the past two days helping them learn how to use MATLAB. If anyone is interested in the basics, I do share that kind of info on my blog here for example, which was a whole five posts on MATLAB basics. It was a rough intro, but an intro nonetheless.

Saturday I spent a little over an hour going over the basics and we stopped before ever really learning how to load the data, much less process the data. Unfortunately, I did not record the meeting because I could not, so Sunday I spent the day painstakingly recording and redoing the basics for them specifically so they could learn. In the background I’ve managed to get all the data converted to the proper format, organized it the way I like to have my data organized, modified a bunch of code that had some things that annoyed me, and heavily added comments to the code I wrote to load the data so the people helping me would understand what I’m doing.

I’ve also started re-writing my pre-processing pipeline with a ton more comments for the people who are helping me so I can walk them through just how to do pre-processing and why it’s important. The pre-processing step is the step where we take our data, filter it and make it all nice and “clean.” It’s arguably the most important step we can do and it’s also a lot of learning, especially for two people who have very little or no programming experience and no MATLAB experience to speak of whatsoever.

The short version of this story is that things are taking a lot longer because I want to do a good job teaching these two the basics. The stuff I wish I was taught. The stuff I did wrong and had to learn how to do right. I want to make sure they don’t make the same mistakes I made so they can go off and make unique and new mistakes on their own. Because mistakes will be made, but I want those mistakes to be something different from the common ones we make.

This means things are going to take me twice as long as they would if I were just doing things on my own. At least twice as long. I don’t mind too much because it forces me to thoroughly document the code I write and that will help future students in the long run, but it does mean that I will need to set even more time aside for my own, uninterrupted, data processing time. Like it or not, I have deadlines to make, the DARPA meeting is coming and I need to be ready for it, so I won’t be teaching them as I’m processing the data like I had originally planned, instead I’m going to do my work, then redo some of it to teach them.

It’s a duplication of effort, but it’s the only (and frankly best way) to get this done on time. Things are getting particularly busy on the hospital side, so I don’t have time to waste. Plus I’m already months behind schedule with my original timeline from my proposal, so any and all time I can recover by speeding up the process I need to do. Until the deadline for graduation comes and passes, I am going to keep working like I will graduate on time, because I really, really, want to graduate in the spring of next year.

The work continues I guess!