3.3 OpenLMIS Reporting Demo

Demo Data

  • DISC Indicators: Full-stock Availability, Closed Vial Wastage, Stocked According to Plan, Functional Status of ColdChain and Forecasted Demand Ratio
  • Country: Mozambique
  • Geographic Scope: Entire country
  • Dates: Clay Crosby (Unlicensed) should we aim for the last 12 months? Clay Crosby (Unlicensed)'s response Mary Jo Kochendorfer (Deactivated): I think this would be a good default in the real world. However, demo data currently spans September - November (3 months), and so most of the slices are hard-coded to that effect currently. We can adjust based on the demo data we generate.
  • Details on the demo data parameters are here.
  • CCE details for Moz are in the excel file PRE-MATCH_CCEI_Database, please ask Mary Jo Kochendorfer (Deactivated) if you need access.

Risk Based Approach to Analysis

ProcessIdeal ViewNotes
  1. What is happening?
visual / at a glance should illuminate the RISK right away
2. What data is driving that state?Drill-down ability from step above
3. Where should I focus?

Ideas:

  • 3-month view of the 'repeat offenders' 
  • show related indicators at that site

4. How can I improve that? What questions do I ask?

static , suggested RCA questions to ask at the bottom

Vidya Sampath to provide the RCA probes for each indicator

New: Mary Jo Kochendorfer (Deactivated) - I thought about this after our meeting – but another cool option to have in this dashboard is note-keeping – in that the user is able to jot down the top 2-3 changes that her/his team has decided to implement as a result of data review and analysis from this dashboard. Would it be possible for the dashboard to have a notes section which also gives people ability to put some dates and "@" tags for tasks to other users?

Clay Crosby (Unlicensed)'s response: not currently available, but I'm flagging this as an enhancement - some kind of messaging/communication functionality within Superset, and an ability to share. At the moment the best way to do this would be via email, Slack, or some external messaging system. 

Positive Performance Approach

Vidya Sampath to insert details on what defines 'good performance'.

Star Performers / Soccer League Style idea moving from relegation to premier to back

Separate dashboard for this; super nascent idea stage ; to be updated 

Full stock availability 

Clay Crosby (Unlicensed) and Vidya Sampath I've created this page to track the discussions on how best to visualize the FSA indicator

Important notes about full stock availability

  • Real value comes in looking at the indicator by facility types (all health facilities or all district stores or all regional stores etc).
ProcessIdeal ViewIdeaNotes
  1. What is happening?
visual / at a glance should illuminate the RISK right away

Idea: Start with a Tree Map View

Groupings for the Tree Map View

  • FSA "Yes" grouping only is yes if all the selected orderables are available during the selected time period at the chosen facility(ies) - label this grouping "Vx Available" in the tree map view.
  • FSA "No" grouping is if any orderable selected is stocked out during the selected time period at the chosen facility(ies) - label this grouping "Stock-outs" in the tree map view.
  • FSA "No Data" grouping if any orderable selected does not have data during the selected time period at the chosen facility(ies) - label this grouping "No Data" in the tree map view.

Filter the Tree Map view by

  • Facility type
  • by commodity type (antigen)
  • by time period

Idea:

There is too much noise if a user selects several antigens and so the bulk of the tree map ends up showing "No" and "No Data" and over-plays the risk as a result. One way to work around this is to display either a simple bar chart which shows antigens with the highest number of sites with no stock or a bubble plot showing 3 variables (antigen, # of sites, # of days out of stock)


2. What data is driving that state?Drill-down ability from step above

When you click on a box in the tree map in Step 1 above, it shows you the data that is contributing to the box. This data can be visualized in the form of a geo-spatial map. Let's say you selected the "No/Stockouts" box in the tree map in Step 1. When you select it, it then shows you a map that highlights the stocked out areas that contributed to the "No" box for the tree map.

Key is to be able to see the 'worst offenders' over time by seeing the drill-down for the selected time period, and the visualization for the two previous time periods to pull out patterns of performance. This gives the user the opportunity  in Step 3 below to "Focus" in further.


3. Where should I focus?

Ideas:

  • A user selects the site from Step 2 above that has long-standing issues
  • show related indicators at that site

When at the site view:

Show the following:

  • CVW
  • SATP
  • See consumption (example below)

  • FSA of the supplying site for the most problematic facilities (for the same time-period and further back)
  • Pull in other indicators for that site (Stock according to Plan, CVW)
  • Re-order information with use information

4. How can I improve that? What questions do I ask? What corrective actions should we take?

static , suggested RCA questions to ask at the bottom

Vidya Sampath to provide the RCA probes and action suggestions for each indicator

Static questions for FSA RCA:

  • Is there a link between availability of vaccines and supplies at this site and the parent site(s)?
  • If yes, what can I do about the re-order practices between this site and the parent site? Is the order request calculated accurately? Is the order fulfilled as requested? Were deliveries or pick-ups delayed often?
  • What is the impact of low availability at the national and regional stores on sites lower down the health system?
  • Are certain sites full availability lower (or higher) than expected every time? If yes, what changes can we make to the amount supplied each order? By what percentage should that amount increase (or decrease)?
  • If we disaggregate by location, are rural (or urban) facilities more stocked out than the other? What can we do about that?
  • Is it specific antigens that seem to be stocked out more than others? What's the reason for that - increased consumption? Mass wastage due to CCE failure? Upstream low stock? Global supply shortage?

Hand written notes


Closed Vial Wastage

Scetches from Vidya Sampath. FYI Clay Crosby (Unlicensed)




Discussion on Visualizations

Status as of  

Will we be able to display Months of Stock (MOS)? Clay Crosby (Unlicensed)'s repsones. We won't have MOS for the week of  but we can aim to have it in for GAVI.

Vidya Sampath and Mary Jo Kochendorfer (Deactivated) please take a look at the new Stock Monitoring dashboard: https://superset.ona.io/superset/dashboard/26/?preselect_filters=%7B%22224%22%3A%7B%22__from%22%3A%22%22%7D%7D
and CVW dashboard: https://superset.ona.io/superset/dashboard/24/?preselect_filters=%7B%7D

Tree map information: starting at the district level so you can show how you can drill into them, although for this initial round of demos obv we will just have data for cuamba (plus the little bit we havein there currently for lichinga)

Geographic Maps: one to show FSA by district, and one to show submissions. currently the map we have is province-levle but our team should be replacing that early next week. if you want to build another map viz, it;'s really easy. create a new slice and configure it like this, the only difference would be changing the metric. when our team creates the new map file, the "ISO 3166..." field should be "district", not "region code" (as we will be creating a district-level geojson)

Full Stock Availability: 

  1. that goes from district -> facility -> commodity type. it excludes CTs where FSA = 0, but that would mean the soh for that CT = 0 for thelast year, which i dont think would happen in the real world
  2. FSA - yes - this does not group by CT, per vidya's instructions that "ALL" orderables have FSA=100 to appear. So, there is no data for this chart currently due to the sparse demo data we have on hand. I expect this to change after craig and the team load data for the cuamba district
  3. FSA - No - this does not group by CT for the same reasons as 2), and for the same reason as 1) (that things with FSA =0 are excluded), this chart is null. again, forthe same reasons as 1), i don't think this would happen in the real world and i expect it to be improved inthe demo data

if the fact that facilities where FSA = 0 for all CTs across the reporting period are excluded is an issue, we would really just want to report on facilitieswhere soh = 0, which can be done, but that's not how i'm interpreting the ask


OpenLMIS: the global initiative for powerful LMIS software