Notes
Writing
Learning in public — building data skills for drug safety, one project at a time. Mostly: how a safety signal is told apart from noise, what the data quietly gets wrong, and what I got wrong on the way.
Subscribe on Substack →Latest
I turned my data project into a tool — and deploying it taught me more than building it
Taking a personal analysis script and building it into a Streamlit tool anyone can use.
Read on Substack ↗- Most-reported isn't most-dangerous — what the drugs themselves reveal Why the most frequently reported drugs in FAERS aren't the ones with the worst safety signals.
- What 1.2 million reports of "drug ineffective" taught me about reading safety data Digging into the most common reaction in FDA adverse-event data and what it actually tells us.
Deep dives
Every project is written up in full — the question, what I found, and where the method broke. These are the longest things I've written.
- PV Intake Assistant Triage F1 0.79 · SNOMED coding 0.54 — precision over coverage.
- Causal Safety: Does an NSAID Cause Kidney Injury? Crude OR 1.19 → matched 2.05 → SCCS 2.14, against a sealed true OR of 2.20.
- ADR Reader
- EHR Survival Analysis in R