A 200-patient synthetic table stayed fully unique after removing every identifier
MaziyarPanahi · x · 2026-07-28
A synthetic patient table can still be uniquely identified even after removing every explicit identifier:
- In a 200-patient dataset, age, sex, ZIP code, and admission date alone produced 74,550 possible combinations.
- The result was that all 200 records remained unique, showing that de-identification is often far weaker than it looks.
- OpenMed 2.0 reportedly reduced the risk to k=15, then still refused to call the table safe.
- The post frames this as a question of whether anyone would sign off on such a dataset.
Related event: Synthetic Data Remains Identifiable Even After Removing Identifiers(2 posts)→
More from Research
- StateAct argues computer-use agents should model program state before pixels — _akhaliq · 2026-07-28
- Cursor details an agent swarm that kept context small and hit 80% on a SQLite-from-docs benchmark — bibryam · 2026-07-28
- Physical AI could make industrial robots easier to deploy — Responsible-Grass452 · 2026-07-28
- Topo, the 1983 programmable home robot that ran code before sensing the world — tobowers · 2026-07-28
- Papers with Code’s robotics hub tracks the field’s main benchmarks and trends — NielsRogge · 2026-07-28
- A chronological map of looped-transformer papers, from Universal Transformers to DeepLoop — KyeGomezB · 2026-07-28