This repository holds the lists of anchors used the paper:

Electronic Medical Record Phenotyping using the Anchor & Learn Framework

by Halpern, Horng, Choi, Sontag 2015

For each condition there is a file that contains a list of anchors in the format:

source|identifier

possible sources are:

• ICD9 billing codes (icd9)
• medication history (med history)
• dispensed medications (med dispensed)
• MD comments (mdcomments)
• Triage assessment (triage)

To translate the identifiers to human readable names, we suggest that users download the following reference files:

• CMS32_DESC_LONG_DX.txt -- download here
• RXNCONSO.RRF -- available as part of the RXNORM package. More info here. Download page. Requires a license which can be requested here.

The medication identifiers are GSN codes from First Databank (source name NDDF in RxNorm), which are only availble in the full releases of RxNorm.

The python script display.py uses the above reference files to add human readable names to every anchor.

Usage: python display.py condition_file