I love visualizing data. And it feels even better if to do this I have to go through a circle of hell called “processing of biomedical texts”. So is the case here.
It is well known that genetic and symptomatic similarities between diseases allow repurposing known drugs for new indications. But what if similarities can also be discovered from the expert knowledge used in planning and conducting clinical trials? The logic is simple: if a lot of different drugs were tested on the same two diseases, it means that these diseases have a strong biological relationship between each other. Let’s see if it’s true.
I decided to visualize the relationships between different types of cancer using research data from ClinicalTrials.gov. The latter has a lot of useful information on publicly and privately supported clinical studies, and each interventional study record contains indications and interventions (e.g., drug or device) received by participants.

#network-graph #cancer #artificial-intelligence #clinical-trials #naturallanguageprocessing