Data: the antagonist in Public Health

It’s one thing for a scientist to come up with a research question. It’s another thing to actually begin attempting to address it. For those of you who have not read my post “A Year Ago,”—you should!—I’ve copied my research question down below. 

How can we expedite the accurate identification of bacterial infections acquired internationally and match patients to the right antibiotic?

In the fall of 2025, I started to brainstorm ideas. I knew that I did not have easy access to a laboratory setting, so I thought about other ways to answer this question. It was then that I came up with the idea of training a Machine Learning model that would help doctors make more accurate and efficient diagnoses of global pathogens by predicting potential outbreaks based on past travel and symptom data. I would train this model on international outbreak and patient data, specifically from India as that felt the most significant due to my personal history. So there, that was easy! The first step was to find data. Unfortunately, I did not anticipate how large of a roadblock this was going to be. 

After speaking to a few professionals at the New York Department of Health and some Epidemiologists, I was directed to a few databases. These included the World Health Organization, Center for Disease Control, and Center for Infectious Disease Research and Policy data. I also looked at some Migration patterns from India to the United States over the past few decades. There were two main issues. Firstly, a lot of these databases did not provide any travel data or any records of patient diagnoses. This data may exist but is likely very confidential and not easily accessible to a high school student. Secondly, even the data on past outbreaks were extremely scattered. Reports were not being consistently released or updated, nor did they give detailed information. I could tell that there was no clear form of global disease communication that currently existed, and found that a bit disheartening. 

If I was a medical professional, it would be super helpful to have global travel and outbreak information native to my intake systems and as part of a patient questionnaire. This would help me be much better informed about types of diseases likely coming from particular cities and countries as well as a lot more prepared to treat them. Not only did I now have to reconcile with my lack of data, I had also figured out that doctors are really not given enough resources to gain knowledge about disease transmission outside of the country they work in. How was I going to train a model on data and knowledge that did not exist?