TO check - NOT AN ISSUE

[idrissait]

DONE - Adding name labels (and maybe data labels) to the map, on the screen display and when it is exported/pasted already included #5

TODO - Show the weekly incidence rate in tables and/or graph for weekly incidence already included

TODO - Get an error box when try to save (saying ‘the file may contain features that are not compatible with CSV). Then the incidence rate is not calculated. switch to xslx?

DONE - The Epi Weeks labels on graphs are too small to read on the screen one per month displayed #38

DONE - Maps –can there be an export button – at the moment have to download ‘Screen snip’ software to copy the image on the screen then paste into Paint and import into Word or PowerPoint – this is putting people off (the data manager in S Kivu is using this system but non-MSF people using the dashboard in SL are put off) already an issue #5

DONE - Can there be an option to have graphs as line graphs instead of bar charts – in Sierra Leone they only want line graphs so are not using the dashboard graphs, same request form S Kivu -> give choice #39

[idrissait]

1) How are the incidence data calculated? I was developing an Epi Bulletin (to be continued by next epi) and including incidence data. Upon doing the calculations and using different years of population projections for the denominator, I could not recreate the same results as those presented on Dashboard. What denominator was used in the calculation? We currently have 2016 population projections for Tonkolili and each PHU. As MSF and the DHMT are using these population projections to direct all programs/activities, is it possible to adjust the Dashboard to use the 2016 projections so that we can also use the incidence data from Dashboard (currently, I was doing my own calculations to produce the incidence data with 2016 population projections but it would be great and much simpler to use the Dashboard - plus the DHMT would likely not be able to do their own incidence calculations every week so the bulletin would not be a transferable product). Note that in the DHMT Excel file for IDSR weekly data, the population is listed as the same (I believe 9999) for each PHU and the correct population projections are not within the IDSR file - all incidence data for each PHU are calculated with the same number!!! (9999)

Incidence calculation is explained in the user documentation (p23) available here: https://github.com/MSF-UK/MSF-Dashboard/blob/master/cfg_dhmt-tonkolili/doc-user/dhmt-tonkolili_doc-user_draft.pdf

IR = Number of cases over the selected weeks / [Population in the area or catchment x Number of weeks] 

In the early version that has been deployed to the field the calulation might give errors when time dimension is filtered by years. This has been fixed in the last release available here: https://github.com/MSF-UK/MSF-Dashboard/releases/tag/v1.0.0_dhmt-tonkolili Some supervision for deployment of a new version is likely to be needed...  

2) Once the incidence rates are corrected, is it possible to also be able to create a line list on clipboard to copy the rates to an Excel file? Currently, we can only get a line list on #cases/deaths

Adding this functionality has been pointed out in some opportunities already. It has been reported, along with maps and chart export functions, here: https://github.com/MSF-UK/MSF-Dashboard/issues/5 This would require some extra development. Exporting the aggregated figures as well as incidence and completeness rates is probably the easiest one though. Probably 2 days of work.  

3) Is it possible to be able to add a feature where the PHU can be labeled? For example, if there is a PHU with a high number of cases, we can attach a label to identify the PHU by name

It requires some developpment as well. A librairy exist to add labels: https://github.com/Leaflet/Leaflet.label it is most likely feasible. Then we could add a button to manually display/hide labels. Probably 3 days of work.  

4) Currently the maps and charts produced by Dashboard are not readable on black and white printouts. This is challenging if we want to produce reports using the Dashboard tool. Is it possible to change the colour coding to patterns so that the charts and maps can be added to reports that will be printed in black and white?

Here it is a difficult topic. I join an example, I do not feel it looks too bad. Then when photocopying, the difference between the tones is more difficulte to notice which can be a problems for maps (but for charts?). One option for the maps would be to have an option to display less categories (3 or 4 instead of 5) so the contrast would be more importante. Probably 1/2 day of work, for tests etc.  

5) Currently when you import the excel files you get errors within 5 cases and 6 deaths. You need to go into the input file and sort the data to delete the 11 errors and then re-import the file. This has to be done every week as the errors date back to much earlier weeks. The DHMT were trained on how to remove these errors but this is something that should be rectified as the Dashboard does not work correctly if errors exist, and the errors are occurring during  the manual data entry at the DHMT. The only way to permanently correct these errors is to go back to all the IDSR week Excel files and correct them within those files, and re-install them within the Dashboard error-free. This would also mean that the DHMT should be given the corrected files, and that there should be re-entry of the files into their Dashboard tools. Furthermore, moving forward, the files need to be cleaned before entry which would be difficult as the files are large and there is a big potential for human error in data entry. Is there a possibility for Dashboard to recognize the common errors and ignore or delete them? If not, it is important that previous files are cleaned (for the 11 errors) and that we continue with this practice moving forward (and include it in the training).

I am not sure I get it right. In principle if you change a value in the database, your changes should remain. Each week you should only import the last week to the database. So if you corrected values of earlier weeks you should not have to correct them again. If you do have to make the changes evey week this something is going wrong (either a bug or a problem in the process). We can do some tests together if needed.

About data errors, in the last release, the dashboard should not fail to load, but we do not and we have not planned to do any automated data cleaning. I think that if errors are detected, the IDSR spreadsheet should be cleaned, and a clean version should be sent to the higher level. We should not let the error propagate nor give the impression that data quality is not an important topic.

---

So the changes to the SL that Bruno says would need development time are:

• Export of incidence rate data in an excel file (2 days)
• Ability to add labels to the PHUs on the dashboard map (3 days)
• Change black and white maps to have 3 or 4 shade categories, not 5 or 6, so that they are more easily to interpret in black and white (1/2 day)

I would say we need to email Idriss again to say that these changes need to be made and to ask if MSF should get Bruno to do this (ir wait for Idriss to return so he can make the changes)?  And we need to clarify with Idriss if MSF has to pay Bruno for this? – So I have copied Idriss in.   @Idriss:  Let us know!

Are you happy with the other 2 answers/suggestions, which don’t require software development work?:

• Incidence rates (the user guidance document Bruno referred to says, on page 9: What happen with population data? Population data is not currently part of the IDSR reporting (all values to 9999). An external population data file is currently used directly in the Dashboard, which should be updated as new figures become available. The current implementation does not allow to take into account changing population figures which would be necessary in some contexts or if the Dashboard happened to be used during a long period of time.)

So it is known that the DHMT files only say 9999 for all populations. The dashboard instead uses population data from an ‘external population data’ file linked to the dashboard. You should be able to find this file and check the population data currently in it, and change the population data if it is incorrect. It also seems that all users need to upgrade to the latest release of the dashboard, at that URL that Bruno gave.  

• Errors in the excel files when they are imported.  It sounds like there might be a mistake in how data is imported each week:  Bruno says “Each week you should only import the last week to the database. So if you corrected values of earlier weeks you should not have to correct them again”; but it sounds like all the data from all past weeks is being imported each time (along with these 11 errors).  Do you think this is the problem? If so, the people/person who import the new data each week need to have this explained to them.   If not, it sounds like you (now that Jordan is gone) need to have more of a conversation with Bruno so that he can fully understand the situation and suggest what change needs to be made – it sounds like all users may need to have new, clean versions of the spreadsheet sent to them to use instead.

[idrissait]

1.       The screen/scale/ data is far too small to be able to be read, to the point of being unusable  on a laptop – this is true for the graph with the epi weeks being far too small to read 2.       The color scale projects as brown on the low quality projectors that are used in the mission .. there needs to be some form of greater contrast as everything looks red/pale red in bad light on a bad screen 3.       There needs to be a button similar to that used on the current MSF data tools, to enable graphs to be copied and pasted in one action, rather than using the snapshot windows tool which needs to be installed on each computer 4.       The issue of cleaning up the data files is still a major one (and an obstacle to allowing the DHMT staff to use the tool quickly) – the EPI CHO has to clean up the data base – the DHMT do not have the time to do this manually every week 5.       It would be useful to have the data presented as a linear trend as well as by case load in the  bar charts (especially for incidence rates)

[idrissait]

I have the same problem as Katanga and I add manually the name for Health zone, but it will be good to have it insert automatically when we encoder the data. My second enquiry is like said N***** and Sierra Leone, I finish to insert the health zone population but when I want to save, this message error appear (in attach) and the incidence rate is not calculate, your comment, please.

For other, when I finish to put all data ( for two last year), I will come back to you for more questions.

1. I entered the 2014 data, 2015 and 2016, is it possible to view data by week and year desired as is done only when you want to view the data by year. 2015 for example? 2. In the graph: why you did not want to present the graphic as a curve instead of histogram?

[idrissait]

DONE - Adding name labels (and maybe data labels) to the map, on the screen display and when it is exported/pasted already included #5 DONE - Show the weekly incidence rate in tables and/or graph for weekly incidence linked with #35 ?

[idrissait]

Difficile d’afficher l’incidence malgré que les données de population par zone de santé y soient intégrées,

Pour les semaines épidémiologiques, est-ce possible de les afficher par années comme on le fait quand on visualiser les données par années (là nous avons la possibilité de trier l’année souhaitée)

Est-ce possible que les noms de zones de santé s’afficher automatiquement quand les données sont saisies ?