2018-06-13 Weekly Check-in

Date

13 Jun 2018

Attendees and/or Technical Stakeholders

• Craig Appl (Unlicensed)
• Clay Crosby (Unlicensed)
• Antonate Maritim (Unlicensed)
• Wambere (Unlicensed)
• Felimone Amone Junior
• Timoteo Chaluco (Unlicensed)
• Dércio Duvane
• Christine Lenihan

• Ben Leibert

Discussion items

DHIS-2 and Coverage Rates

Question: DHIS-2 returns information for Polio 0, 1, 2, and 3, and similarly numbers other vaccines. How should we interpret this?

Answer: The vaccine for Polio and other diseases is sometimes given in a series. The antigen is the same, and is simply administered every x weeks. The Dados_Coberturas tab in the existing Dashboard includes "Polio3," which refers to the third and final dispensation of the antigen that's necessary. Meanwhile, the Dados_IQV tab lists "Polio1 e 3," which refers to doses 1 through 3 of the same vaccine.

As mentioned, the Dados_Coberturas tab shows just the last dose of Polio administered. After someone has received it, they've completed the regimen and may be considered "covered." Even though the Dashboard may not always need to distinguish between the different doses of something like Polio that have been administered, we shouldn't throw data away. Therefore, if DHIS-2 distinguishes between the various doses, we should too when importing its data. We may not show the differences in the UI now, but may need to in the future.

Regarding the Coverage and Cov columns on the Dados_IQV tab: a coverage of 0% means that the same number of children who were previously vaccinated returned for their requisite follow-up. A coverage of 14 percent, for example, means that you're 14% over this goal because more children arrived for follow-up treatments than expected. This might happen if they had their initial treatment in a neighboring district. Finally, a negative value of -14% would mean that 14% fewer children arrived for follow-up vaccinations than hoped and expected.

The "Cov" (short for "coverage") column in Dados_IQV is hardcoded. The Vaccines_Needs_Formula_Stock Availability_Months_weeks_Days.xlsx file describes to calculate it.

DHIS-2 and Brigades

Question: DHIS-2 distinguishes between facilities and mobile-brigades, as well as between male and female. Do we need to do the same?

Answer: We may need to in the future and thus shouldn't throw away data. Just as DHIS-2 distinguishes between them, so to should we while persisting its data. Nevertheless, for now, Superset's dashboards don't need to distinguish between them. The materialized view that it references can thus add male and female related values together - and do the same with facility and mobile-brigades. For performance reasons, this addition should occur during the data-ingestion phase rather than only the fly as users access the dashboard. 

Format of Email Attachments

We'll plan to use .xls rather than .csv files because this will allow us to provide templates which guide users toward entering properly formatted data. This could avoid confusion, for example, about when to quote certain types of input and when not to. 

Our users rely exclusively on Excel - but on an eclectic mix of versions. We have to support use of Excel 2007, which means we can't require use of the newer .xlsx format. Meanwhile, because newer versions of Excel encourage users to rely on the .xlsx format, a process requiring the opposite would be error prone. We'll therefore aim to support both .xls and .xlsx formats. If this proves impossible, we'll have the system notify users via email if their submissions are delivered in an unexpected format.

Translations

The latest Brazilian Portuguese translations in Superset are somewhat lacking. Meanwhile, however, the Continental Portuguese ones in this pull request look great. We should thus ask the Superset team to merge the submission. 

Historical Data

The fact that we don't have historical data to test with will become a pain point very soon. Unfortunately, though, the folks at the MoH have been at workshops and unable to provide us with the data.

Question: Is the e-SISATFE data autogenerated?

Answer: Their system generates an Excel spreadsheet with lots of data. A user, in turn, will manually extract and send just the part of it they wish to share. Although the process therefore involves some manual effort, nobody is keying in the data by hand.  

Progress, Risks, and Schedule

We've made some good progress on the NexLeaf work. Their API is in flux, however, which represents a minor risk to us.

Another risk is that Dercio will be traveling next week. We've relied on him to answer a variety of questions, and will really need Timoteo to be available in his stead. To aid with communication, Ben will thus reach out via WhatsApp.

We now have the latest version of Superset server running at dashboard-pav.openlmis.org and accessible via SSL. NiFi is setup in a similar manner.

We're making reasonable progress on the NiFi connectors as well.

Friday, 6/15, is a holiday in Kenya. Meanwhile, the Monday of 6/25 is one in Mozambique.

We hope to complete the Dashboard by Tuesday, 6/26, and demo it to the team then. This will give us several days prior to the end of the month to incorporate feedback.

Action items

•  Ben Leibert or Clay Crosby (Unlicensed) will ask the Superset team to merge this pull request.