Implementation Notes

• 16 MB limit in MongoDB for commands (array)
• A schema should be implemented in the Mongo database
• Modeling encounters - Encounters are nested within encounters, which create a master encounter (i.e."Master encounter (clinic visit)" contains "visit with nurse (vital signs) encounter" and "visit with clinician encounter")
• Two Basic Cases for queries: Offline analytics batch queries that can be done overnight and Queries on a specific patient
• The classifier will return an array of integers (for class-based queries), which we will then “pass in” to the database
• We will have to rely on numerical identifiers to query in the database
• Queries will be against the coded post-coordinated expressions; uncoded post-coordinated expressions don’t need to be housed in the Mongo database
• Cardinality for observations in encounters has an upper bound of 100-200
• Prototype should be able to be used in multiple CDS systems