Research Journal: Oct 16 - Oct 22

Where there's data there's a way

October 16

I’m still on wedding brain right now, and I barely got enough sleep because I shared a room with my parents. Taking this day as a day to catch up on personal chores.

October 17

Today I flew back to San Diego, so I didn’t have much motivation to get any research done. I did catch up on a lot of administrative duties that I had been collecting, so that was productive at least.

October 18

Bad news today: the data that we have from the Just Walk Trial doesn’t seem to be a good option to use with the MEBN. There’s a bug in the currently implemented code, so I need to see where that’s coming in.

I think I might have to ask Sonia if she has access to an alternative dataset for us to use. I’ll need to delve deeper into each of the models to get a better understanding on how Scutari’s code works.

October 19

Today was a big day: I have a second laptop to run my simulations on. Not only that, the new laptop is a lot faster than my own, so the pace that these simulations will be generated will go so much faster.

Sonia recommended that I look through different N-of-1 trials to try to demonstrate that our method was effective. I’ll need to read though a bunch of manuscripts, and I bet that I’ll have to do a bunch of emailing, but I guess it’s a part of the process.

Moderna is having a biostatistics internship, so I was definitely interested in that. There wasn’t much to the application, so I can only hope that my skillset is what they want.

October 20

Brought back the writing habit that I had a bit earlier: a mandatory writing period everyday. If I contribute to all the projects I have in a day, even just a little bit, I will tell myself I can’t feel guilty. I think that will help on the mental health front.

I need data! I sent out emails to a few people to ask them for their N-of-1 data, so hopefully I get some responses. I’m not sure how many PhD students reach out to strangers to ask for data, but hopefully the N-of-1 community is accommodating.

October 21

Break day today! I feel a little guilty here because I didn’t get my 3 major tasks done, but I think it would be more fair to allocate this time to the other days of the week to offset this day where I know that Yuwei and I will just be hanging out.

October 22

Delved into the code today to figure out why it wasn’t working. It turns out the original Scutari paper assumed that all nodes in the network needed to have the grouping variable as a parent. In the case of an N-of-1 trial, we know that the independent variables are controlled by the investigators, so it shouldn’t be different between people. I’ll change the implementation so that this is accounted for. This should remove all of the model singularities and give me a more complete parameterization.