Q1: What is the relationship between SHRIMP and SHRINE
A1: SHRIMP was used to map SHRINE vocabularies across the various Harvard Teaching Hospitals. SHRIMP could be used for other purposes.
Q2: Why is SHRIMP designed for simple "association mappings" instead of domain specific ontologies like Pathology or Imaging?
A2: The primary use case for SHRIMP is to create a common search strategy for related terms. To create ontologies that are specific to a research area, we recommend evaluating the i2b2 Ontology and the Bioontology
Has the scrubber actually been used to share free-text patient data?
Yes. The scrubber was used to de-identify over 1 million pathology reports across Harvard affiliated teaching hospitals.
Will my hospital IRB approve using this Scrubber?
This will depend on the expected stringency and intended use of the de-identified data. At Harvard, the scrubber was IRB approved for full-text searching in pathology reports. In the Harvard implementations, up to 400 reports could be shared with an approved investigator.
How well does the scrubber work?
See publications for an in depth manual review of cases.
Recent independent reviews at 2 large HMO sites suggest even better performance since the 2006 report.
Updated report forthcoming.
Does the scrubber also help find medical concepts? (autocoding)
This is the top priority for 2011, see project roadmap.
Since the scrubber can find phrases of any kind, this is definitely possible, even in the existing released code.