Stream: implementers
Topic: Resource(s) to represent output from ML algorithms
Sathish Tiptur (Sep 20 2019 at 08:46):
Are there any resources or any work happening in the area of defining FHIR resources to represent output from Machine Learning algorithms? For example, a machine learning algorithm analyzes various vitals and generates a risk score with probability. Any suggestions to represent such information as FHIR resources is much appreciated. Thanks!
Lloyd McKenzie (Sep 20 2019 at 12:56):
How you represent the output would depend on what the ML is determining. For the specific case you identified, either RiskAssessment or Observation would be appropriate.
Sathish Tiptur (Sep 20 2019 at 13:04):
Thanks @Lloyd McKenzie
Morten Ernebjerg (Sep 23 2019 at 07:42):
A while back, there was also a discussion of how to represent confidence for data extracted/analyzed by algorithms here: https://chat.fhir.org/#narrow/stream/179312-data-extraction.20services/topic/confidence-range.20extension Not quite sure what happened to this, @Simone Heckmann is there now an extension for this?
Last updated: Apr 12 2022 at 19:14 UTC
