Skip to main content

Clinical data & AI

Clinical data & AI

Interoperability with bedside devices and machine learning over imaging archives.

Clinical AI is only as good as the data it can reach. In a hospital, the most valuable data is locked behind proprietary monitoring hardware, varied DICOM vocabularies, and patient-privacy controls that — rightly — make casual access impossible.

This theme covers MESH|Lab’s work on getting clinical data into a form research and AI workflows can use, and on the models we then train over it.

What we build

  • Interoperability tools for bedside monitoring devices — most recently the Nihon Kohden monitoring family, where we extract waveform data into open, structured formats for retrospective research.

Why it matters

The bottleneck for clinical AI is rarely the model — it is access to data that reflects the clinical reality the model will eventually be evaluated on. By building the access pipelines openly and validating them against the workflows the data was generated by, we make the downstream science more reproducible.

See the Projects page for the current portfolio under this theme.