Antimicrobial Resistance (AMR) Data Analysis

Antimicrobial resistance is one of the harder problems in public health: the bacteria that antibiotics used to kill are increasingly surviving them. Understanding where that is happening requires surveillance data from hospitals — and that data arrives messy.

This is my ongoing work at the Emerging Technologies for Health (ETH) Research and Development Laboratory at MUHAS, analysing AMR data to find patterns across pathogens, specimens and susceptibility results, and turning raw hospital records into datasets that can actually be modelled.

Most of it is the unglamorous half of data science, and it is the half that decides whether any conclusion drawn later is trustworthy.

Hospital names spelled six different ways. Ages recorded as text. Specimen dates in inconsistent formats. Regions and districts that do not match the official list. Every one of those has to be resolved carefully — and every decision documented, because in health research a silent assumption becomes a wrong finding.

I came to data science from 3D art, and the thing that surprised me is how similar the discipline is. In both, the work that nobody sees — clean topology, clean data — is what decides whether the final result holds up.