2020 03 27 Minutes
- Keith Boone
- Christina Caraballo
- Katie Iser
- Dave Pyk
- William Moroz
- John Moehrke
- Ben Hanley
- Dale Evanchak
- Gino Canessa
- Josh Mandel
- Joe Wall
- Hans Buitendijk
- Ankit Agarwal
- Steve Nichols
- Chris Lindop
- Bill Mehegan
- Carl Anderson
- Stacy Kinkead
Review of Security Considerations
We have a good idea of maximal level of classification desired, what's the minimum set?
- One of the things we are hoping this project can deliver is, at the very least, instructions for how to configure a FHIR server to work with the project; receiving data. Works well.
- How difficult is it to add 1 or 2 custom query parameters for Microsoft in Azure?
- Not super difficult
- Queries for 1st story - Find a bed
- 2 Custom Query parameters - one on quantity and one on value (value near operating on location resource for a characteristic that references a location)
- 3rd one adding search capability for a specific period
- Tried to do the same thing with Measure Report but could not get there and it is because we are trying to test 2-3 different things within the same subcomponent where there can be multiple components of the same type.
- Only will be dealing with most recent data, the set of characteristics is associated with a persistent UUID with its own set of characteristics - challenge around maintaining that because it complicates the history
- Getting queries for the physician - pulled from a FHIR server and processed out into different systems (like an Ai step) generally pulling over data, pre-processing it, and having it formatted. Does this play into this at all?
- Our hope is to provide a FHIR server that can provide useful content for someone who has a mapping application but may not know a lot about FHIR
- Would the person go back to the server and do multiple queries?
- Queries for a specific region
- Standing up individual FHIR servers for different regions?
- For piloting this, there would be a FHIR server stood up for a given region and would work with data coming in from that particular region and there is a way to aggregate data up so a system coming behind at a national level will periodically go off and query regional servers and aggregate group resources at a national view
- There is a FHIR server that can receive data and applications written on top of it that don't have to do a lot of work post-processing to produce maps/graphs
- Is this work to define both sides of a value-add intermediary
- Seeing two sides of the same intermediary: the raw poll and the value add output
- Intermediary can also function to do analytics over the data that is going into it, or another component can go over the data to produce a measure report or alerts
- Started sketching this out today
- Output would be simple> physician and 4-5 fields
- Think this would work with Group
- Issues: confidence in having the full data set and varying granularity
- How do we ensure, when a data point is missing, WHY it is missing.
- In this ward I know specific data and at a hospital, I have more general data
- Min vs max characteristics
- If we are including the pre-processing step
- Vote by survey - shared in Ringcentral Chat and Zulip Chat
- Are our goals about reporting data clearly or making a client-facing interface that makes querying as easy as possible.
- Maybe make a decision in principle to work towards one and use the other if we hit a wall
- 7 votes for Measure Report, 2 for Group
- Going between the two is a transform, so if someone wants to take the step to build the other that is not too hard
- How difficult is it to add 1 or 2 custom query parameters for Microsoft in Azure?
Review of Security Considerations
- Business sensitive or anonymous read access
- Do recommend TOS with server-side authorization only and server-side confirmations
- Normal HTTPS server-side
- Might have a client identify and might record that
- Purpose of use
- Use case is based on public health reporting
- Where would PUBHLTH code go? In audit element there is a purpose of use element > put it there
- Will be part of policy agreement on the use of API
- Not covering the use of data at the server or client but suggest considering that
- Use case is based on public health reporting
- Since no PHI want to keep it simple
- Profile AuditEvent to generate a profile
- Profile AuditEvent that details what client would record and what server would record IF they choose to use
- Ongoing opportunities to continue to revise this and make updates but this is a good start
We have a good idea of maximal level of classification desired, what's the minimum set?
- HAvBED2 Bed Classification - work that comes out of original bird flu use case in 2008
- List for Cerner sits closer than headings - left-hand side breakdowns
- Hans to update wiki information
- Ask Michael Donnelly to upload the most current copy of Recommended Bed Reporting Measures to the wiki
- Do we know where ICU classifications came from?
- FHIR CCU - HL7 v3 - Superset in FHIR coding and we are looking at the sub-types
- Hans to check which version Cerner is reporting to
- Want to make it clear that you don't have to report all of these (not all hospitals have burn units or report everything but ED)
- Then people can get into finer levels of detail for age-restriction and purpose of the room that would be okay
- In terms of how does this aggregate up? Some people reporting a certain level of detail and other reporting at the highest level. We should make it clear in systems that are making use of the data. Definitions are helpful for defining the categories that are being reported.
- If using this to aggregate data up, here is the minimum amount of data needed and how to work with deeply and coarsely segmented data.
- Being able to report on trends: may only be able to look at individual facility reports to start trending at a detailed level
- Epic and Cerner are able to produce ICU and hospital
- Don't see ED, though there are standalone emergency department systems that can report ED
- Made things more restrictive to receive some push back- will make those changes to the information
- Wayne Kubick at HL7 says we'll have a project scope statement reviewed by next week, which is record-breaking time. Thanks David!
- David will let Keith transfer repo ownership to HL7 as soon as he's ready. This means the wiki, issues, will all come along.
- Datavant Pandemic Response Hackathon at 4pm today
- US Based includes HHS
- Proposed project dealing with mapping tools around utilization to keep healthcare professionals safe
- No one on-call currently planning on attending
- Keith may attend as a FHIR SME to hear about ideas
Live notes: https://hackmd.io/3pj-dBHNRp6qvYKuw3SnqQ)
TODO(Hans): ../issues/30
Update Bed-Classification with details on Cerner's reporting
TODO(Keith): ../issues/29
Transfer https://github.com/AudaciousInquiry/saner-ig to HL7 github org
Keith - ask Michael Donnelly to upload the most current copy of Recommended Bed Reporting Measures to the wiki.
Remember to Register for HL7 Connectathon May 13-15 and Fill out the Survey
Sign Up to Fight COVID-19 with FHIR 