Better Earlier Than Later

If it's a good idea, somebody's probably already thought of it

It’s the end of my second year in the Ph.D program. That means it’s finally time to transition to full time research, which for a biostatistics program, means a lot of being at home reading and programming. I got a taste of having to find papers on a topic during the Spring Quarter and now my summer break is just doing this… but more.

An AI interpretation to “phd student on summer break relaxing”

What’s been happening since last time?

Getting Pre-Scooped

After spending a few weeks reading the literature on micro-randomized trials, I thought that I had finally thought of an idea that I could use for a paper. This idea was to propose using a new type of model to the trial that would be easier than what’s been currently established.

I thought about this idea for several days, thinking over how the new model might be adjusted to suit the particular needs of the micro-randomized trial. Using environmental information? Just throw it into the covariates. Personalizing the model to a person? It’s doable! Learning the best times to contact a person? That’s essentially built into the model.

I even proposed it to my advisor, and she thought it was a great idea. I started an Overleaf document, along with a bunch of notes. I started to relook at each of the manuscripts I gathered to add them to the introduction.

Then, I found a paper that I knew to be relevant, but didn’t put in a lot of time to read. I think you know where this is going.

I was scanning the document, and I saw my idea being used. Not only that, it felt like they had covered all of the little ideas that I had come up with with a few days of thinking. It felt so good to think of an original idea, only to feel so dumb when I found out that it was not.

The revelation wasn’t totally without its benefits though. I had put in some research into the specific model, the Gaussian Process, and still feel it has some potential for novel use in biostatistic settings. It’s apparently a widely known machine learning trend, which explains why I haven’t really heard about it in my classes. Despite the disappointment, it’s good to realize this after a two weeks of reading, as opposed to two months before I’m supposed to graduate.

The reading continues.

Wranging Citations

One mistake I’ve realized in my early reading was that I was not accurately keeping track of my citations. I was originally just taking notes on each article, but I realized that this was going to be unwieldy for later use. I intended for the notes to be referred back to when I’d be writing an introduction, but I’m definitely not reading through those again.

I opted for a new database approach in Notion. I could write my own notes about the manuscript elsewhere, but I’ve made it a point to convert the citation into Bibtex, along with one-sentence description of why it’s useful. Maybe it might merit adding some bullet points in there if different papers that use it for different reasons.

Full Time Management

With no classes, I have all the time in the day for all of my activities. I’m a person who needs a routine, so incorporating all this reading and writing into my day needs some serious thought. Balancing this with exercise and relaxation time is also a consideration. The next three years of my life are going to be like this, so I should try to take full advantage of it.

Until the next check in!

Somethings I found

• Mini DALL-E: enter a text prompt to the algorithm and get a picture of it! Most of my queries look like beginner paintings. I used this to make all of the pictures in the post.