To go back to the home page, click here.

Why this Project

The Problem

Keeping yourself up-to-date with new scientifics articles in any field is a hard task. While you are still on health training you can easily attend grand rounds or journal clubs to discuss new articles. However, after training is over, you have less time and less people helping you keep you up-to-date.

This is even more critical in oncology (my are of expertise). Cancer is the highest and fastest growing topic in PubMed (link), and consequently being an up-to-date health professional is even harder. Common sources of information for many health professionals include conferences, journals websites, X (ex-Twitter), Bluesky, and industry sponsored meetings.

How People Find Useful Information Around

In my personal experience (in Brazi), people would be reading new publications mostly from Twitter and industry meetings during the year. And this was a problem (in my opinion) because you are exposed to highly biased publications and opinions and I kept myself wondering “what am I missing?”.

A Possible Solution

As someone with a relative good background in programming languages (mostly R, but also Python), I decided to create a personal newsletter with new released scientific articles from PubMed (cancer-related). I subscribed to PubMed notifications center but found it useless (not accurate of what was being published). My next try was creating a R script to query PubMed and send me daily new updates. This worked, but the number of articles was massive (≥ 50 articles a day!) and many were out of topic (despite an optimized query).

A Better Solution

That’s when I thought about using large language models (LLM) in this task. I started using them to filter the daily list of articles as if it were in a systematic review. And finally this worked much better! Since I began doing this in early 2024, the prompt has been optimized and the results were getting better. I started sharing the newsletter with friends in mid-2024 and received an overall positive feedback.

Why a Website

Although newsletters are an awesome way of sharing text and ideas, not everyone likes receiving weekly or bi-weekly emails. In addition, in my friends circle newsletter, only 20-30% (which is not bad actually) opened the e-mails and sometimes I would feel like annoying them. That’s why I decided to share this project in a website (here!). Here you will find the same information I used to share in a newsletter, but with more categories and better reading (email layout is hard to address).

Future Directions

I’m still planning in re-opening a newsletter in the future. If I do, I will be sharing the link to subscription in this website.

About Me

I’m a MD, Ph.D researcher from Brazil, now currently based in Houston, United States for a sub-specialty fellowship in oncology (thoracic and head and neck cancers). I haven’t done any formal teaching in programming, but have been using R (and tidyverse) since 2018 and my Ph.D. was in epidemiology. That being said, I use R or Python in 99% of my current research projects, from clinical to bioinformatics. Recently have been very interested in large language models for health.

If you want to get in touch, please reach me in BlueSky — which I do not use often, but will try my best to respond there.