Stream: implementers
Topic: Standards for AI results, like NLP NER
Sam Schifman (Nov 04 2021 at 14:23):
Hi, My name is Sam Schifman, Chief Architect at Diameter Health, I have been doing some research into AI, particularly around NLP, and have noticed a lack of standards around the results coming from these solutions. For example, an InferredEntity and InferredRelationship resource types could capture a great deal of what these solutions output. Also, adding to the Meta tag we could link resource inferred or supported by InferredEntities to those resources and represent the confidence so that they can be used with full knowledge of their inferred nature. This could greatly improve the interoperability of this data and unlock its potential be used in multiple use cases. I was wondering if anyone else in the FHIR community is already thinking about this and, if so, how I could connect with them?
Vivian Neilley (Jan 31 2022 at 16:29):
Hi Sam! I actually have have been thinking about this a bunch, I just did my Masters thesis on this. Would love to connect!
Josh Mandel (Jan 31 2022 at 16:59):
Some discussion on related work at #fhir/infrastructure-wg > NLP Derived Elements -- we hosted a community call last week that might be worth reviewing (notes + recording are in linked chat).
Brendan Keeler (Feb 03 2022 at 02:01):
(deleted)
Last updated: Apr 12 2022 at 19:14 UTC
