Getting started
Find a professor / head of department in the area that you’re interested in. Email / talk to them (yes, get out of your comfort zone ~ it’s fun, and sets you apart from anybody else who didn’t have the foresight or courage to do it), saying something along the lines of “Hey, I’m super interested in learning more about this field, and have a couple cool machine-learning based tools that can help us improve the efficiency of the research process for the department to use, not just for my paper, but for future papers too. These have been used in previous departments to streamline data extraction and processing, and I’d love to see them in action helping your department. I’m happy to take any topic you see relevant/I have a couple research questions that I’d like to propose, and, pending your approval, I’ll run point on writing the research protocol, doing the data extraction and writing. I’ll do all the grunt work, as I’m really keen to make an impact in this space, all you’ll have to do is provide me with whatever direction you see as being relevant. Will you let me know your thoughts?”
Protocol writing
Use GPT, LatEx and previous example of good research protocol Undergraduate_Research_Protocols (11).pdf
169.9 kB
Put something like this into GPT with the prompt:
“Here is an example of a perfect research protocol, I want to write a similar research protocol on (insert a topic), please write the full research protocol for me but I can submit to reviewers, make it professional neat and very impressive. Include all of the necessary elements for a professional research particle including the Helsinki guidelines.”
Boom.
Research protocol ✅
Just send it to your supervisor for feedback, then put their feedback into GPT with the original and ask it to edit it (unless the feedback warrants more input and fleshing out from your side.)
Data gathering
Available tools
NHLS data extractor - extracts and tabularises any lab data from the NHLS using a web-driver PyWin32 based programs that extract data from any word doc based templates Langchain based programs that use AI to extract and categorise information from long paragraphs into machine-usable databases (e.g classifying types of cancer based on the results typed out in the report section of the NHLS) Use GPT to do data manipulation for you (this will take some time and skill to learn how to prompt it, but the results are well worth it.) Python programs to merge for anything in between
Writing
Do data analysis with GPT Put that analysis, along with the research protocol, back into GPT and ask it to write a research paper for you Send to your supervisor for review Once they're happy with that then start sending to journals, they'll ask you to make edits, put those into GPT. Rinse and repeat until you have a publication