The purpose of this post is to share the current thoughts/plans for the soon-to-launch Analytics Squad.
@akimaina and @bashir and I have reviewed the results of the recent poll in Analytics: How can we make your life easier?, and also a couple follow up messages with a little more detail about peoples’ problems. The main takeaway, in general, is that there’s an acute need for this work!
Before we fully kick off a squad we want to be very clear on some tangible use cases we can address right away, so that the first few weeks of work are ready for people to tackle together, and our community will be interested in trying out the tech we build (hard to successfully launch something if no one is very motivated to try it!). We have also realized that the initial Patient-Level Indicator Reporting work that needs to be done is exactly what the Analytics/ETL squad should help accomplish anyway. @ibacher and @bashir have suggested that the Analytics and PLIR squad should be folded together. Our current thinking is that we should combine and not further separate these efforts.
The idea is that we’ll use some Patient-Level Indicators as the proof-points/tangible use cases for our ETL work. We’ll be formally announcing the squad launch shortly and opening up to people to join this dedicated effort.
Here’s what we’ve been doing the last 2 weeks:
Working space for requirements gathering: In Confluence Here
@akimaina worked with implementers to identify a couple specific PEPFAR indicators which we will use like a canary-test for our ETL work (list below). The list includes simple, moderate, and high-complexity indicators. Theory: if our approach works for all these, we not only have a good sense that our ETL proof of concept worked, but we also have immediate value for the 90%+ of OMRS implementations that need to do PEPFAR MER reporting. (This includes TX_PVLS which is the focus example of the PLIR award. There will be additional PLIR-related coordination work required later, to plug-in to external data collections like the OpenHIE sandbox Jembi is going to work on, but this is a good start.)
PEPFAR MER Aggregates Identified for ETL PoC: (please see the summary table in the requirements sheet for better descriptions)
- TB_ART - Treatment: Proportion on ART and TB+
- TB_PREV - Prevention: Proportion on ART and on TB preventive treatment & completed
- TB_STAT - Testing: % TB cases HIV+
- TX_CURR - Treatment: # on ART
- TX_ML - Treatment: # on ART and no recent contact
- TX_NEW - Treatment: # newly on ART
- TX_PVLS - Viral Suppression: % on ART with suppressed VL
- TX_RTT - Treatment: # on ART with no recent clinical contact
- TX_TB - Treatment: Proportion on ART and start TB treatment
- PEPFAR MER Aggregates Identified for ETL PoC: (please see the summary table in the requirements sheet for better descriptions)
Related (more under the FHIR squad work): @Bashir Sadjad is working on a proof of concept of getting analytics on FHIR to work. TL;DR: Transform OpenMRS data into FHIR resources. It has two parts; one is the streaming part which Bashier recently demoed in one of the FHIR Squad meetings (currently based on the Atom Feed module but can change in future). The other part is batching which is just getting started.
Thoughts and feedback welcome!