 Okay, hi everyone. Welcome again. My name is Ola. I'm from the University of Oslo. I'm the DHS to Implementation Coordinator. In this session we'll present an exciting collaboration between UNICEF, HIST Vietnam and the University of Oslo. So the data exchange on the integration platform, DXP, is an initiative by UNICEF to link global survey data to national DHS2 routine data. UNICEF and UIO, we work together to make the global survey data available on the DHS platform, DHS format, and integrated with the WHO indicators and visualizations from the digital packages that we have talked a lot about this week. HIST Vietnam then developed a custom app to be able to do some of the more advanced triangulation visualizations that we could not do through the standard applications. HIST Vietnam is also supporting a pilot of this project in Laos. Tyler Port from UNICEF's Data and Analytics section will start by presenting the concept and the design, and then John will take over and give us a demo of the cool visualizations and talk about the process of integrating this in Laos. We'll try to make time for some questions at the end. Remember that you have to post your questions in the COP page. You can find the link either in the SCET agenda app or in the Zoom chat that we have here. And then I'll pick a few and we try to respond and then we'll respond to everything in writing on the COP page after this session. So thank you everyone for joining and over to Tyler. All right, thanks Ola. Thanks so much. Thanks everyone for joining as well. And I'd also just like to start the session today from my end by just taking a moment to thank the organizers of the conference. I know it's been a crazy last nine months or so, so I'm always appreciative of how much work goes into pulling off these virtual events. And it's always exciting to see people able to still come together for this. So thanks so much for that. As Ola mentioned, you know, in this session we'll be talking and describing the data exchange and integration platform, which is an effort that we've been working on to facilitate large-scale bi-directional data exchange integration and triangulation, leveraging both global and national-level data systems. And then at the end I'll hand it over to my colleague John, who will walk through a live demo of the current platform. So here's a brief overview. Kind of discuss that so I won't spend any more time on it. I'll just jump right in. So the reason for this effort is the pervasive challenge of fragmentation between data systems. We know and talk a lot about fragmentation within countries, but there are also critical challenges that exist between the global and national level data systems. And importantly, there are some significant opportunities that we could leverage if we could reduce this fragmentation. And when we look at data systems at the global and national levels, we see that data have different primary sources. So at the global level, we're typically talking about household survey data and model estimates and at the national level, really talking about, you know, routine administrative data. And these platforms are often, you know, managed on incompatible platforms and non-standard data structures and definitions. And because of this, it leads to insufficient data sharing both across and within countries and regions, resulting in inadequate data utilization and suboptimal programs and results. And if we look deeper into these data sources, we see that they each have advantages and limitations. But I think what's really important here is that these comparative advantages and limitations are complementary. So if we look at frequency, you know, at the global level systems, you know, for a country, you're really only getting an update every two to three years. You know, at the national level, obviously, you have more frequent, you know, hence the routine data systems. So you're talking about monthly, if not weekly or daily updates, depending on the data. We talk about geographic granularity, you know, at the global level, you know, you're usually only getting one, maybe two levels of subnational disaggregation. Whereas nationally, you have data down to the district level or the facility level or community level and sometimes even individual. But at the global level, you know, we also have data that are sampled for valid coverage estimation. Whereas at the national systems, you have persistent challenges with denominators to be able to come up with, you know, accurate population-based coverage estimates. You know, at the global level, the data are designed to be population-based coverage estimates, whereas you often have facility-based coverage estimates at the national level. And then like at the global level, you're also able to, you know, whereas at the national level, the data you're getting are usually limited to your access, your coverage and your quality, whereas through the global level data systems, the surveys, and you're able to capture additional domains like knowledge and attitudes and practices, access, demand, coverage, and quality. And you also have, you know, you're able to disaggregate your data beyond just the age and sex that you usually have in a national system, which, you know, is inherently limited by the aggregation process. And at the global level data systems, you're able to really disaggregate not only by age and sex, but often by education and wealth, residence, ethnicity, language, and a whole host of other parameters. And so because of this, because of these sort of complementary advantages and limitations, we've found that integrating and triangulating between these two data systems can provide a great opportunity to improve our understanding of health system performance. So this is the ultimate objective behind the data exchange platform, or DXP, as we'll call it here, is to connect these data systems at the global and national levels and integrate and triangulate the data sets. So what is DXP exactly? It's a proof of concept intermediary platform. It's aligned to both the international standard definitions for survey estimates and global models, as well as your routine health facility reported indicators. And it can facilitate bidirectional exchange of data between global and national systems and support the integration and triangulation of those data sets. In terms of the role of UNICEF, we really see ourselves as just proving the technological feasibility, sustainability, and value of this approach. And our plan is to generate evidence to advocate for a more long-term multi-agency solution to data exchange and integration, possibly within the under the umbrella of the Health Data Collaborative or one of the interagency SDG groups or something along those lines. And something that can be more sort of sustainable in the long term to support data quality use and reporting. And just a note as well that we're currently limiting our effort to health data in SDMX ADX standard as our test case, but we do have the intent to show that this is scalable to other sectors and standards. And we've already had some discussions with colleagues from the education sector as well. So what are some of the opportunities we have seen for integrating global data sets into national HMIS? This is important because this is really the primary focus of this effort is how can we get more data to countries in the systems and platforms that they're already using in a way that meets one of their specific data needs? And what functionality does this process allow at the national and subnational level? You can see them listed here, but more nuanced trends and analysis, validation and calibration of routine data and coverage estimates from your routine systems, calculation of smaller area denominators, apportioning national estimates to smaller areas, and estimating service coverage that's not currently captured in HMIS. So for example, if you're not integrating community level data or you don't have good reporting from the private sector, you can offset some of those challenges by pulling in population-based coverage estimates from surveys and using that to sort of triangulate a proper coverage estimate from both sources. And likewise, we're looking at not just pushing data to the countries, but also seeing if we can reverse the flow of data for a select set of government-approved and authorized indicators because we feel that this has the potential to have a greater range of data available at the global level to improve estimation and analysis efforts, but also there's an opportunity here for countries to have greater representation in the global databases while also automating some of the reporting mechanisms for the select administrative data. So essentially working towards more frequent data updates and more data coverage, while at the same time reducing some of the burden on countries and international organizations so that you're not having to go through these manual ad hoc processes to report the same numbers to 12 different partners at the global level, trying where we're having discussions on how we can consolidate and automate some of that to make it a bit more user-friendly for countries and development partners alike. I'll be quick on this slide. So I think an important question is why now? There have been some recent developments at both the global and national levels that have really made this effort possible. Globally, we have the adoption of common data standards at UNICEF and the SDGs and many other partners. So we have this sort of convergence on a common data standard there. We have overhauled and modernized our UNICEF data architecture with unified codes and data structure definitions, which allows us to have greater access to a larger range of data to push to countries. And then also the development and rollout of the indicator standards for the facility reported data and HMIS, which are the WHO UNICEF packages that Ola was referring to and have been mentioned several times throughout this conference. And then nationally, we're looking at alignment to talking about the opportunity that there's gradual alignment towards the HMIS indicator standards, as well as proliferation of DHIS2 and the ADX standard. And this is really important because I think if there had been a decree 10 years ago that every country has to start using a single data standard, I'm not sure how successful that would have been, but because of the success of the DHIS2 program, that has also led to a large scale adoption of a common data standard, which allows us to create solutions that work for many countries from the get-go. So I think that's a huge, huge factor. And then on the technology side, the standard that we've been adopting mostly at the global level, SDMX, and the standard that's being largely adopted in ministries of health through DHIS2 is ADX, our two very complementary standards. And in fact, ADX was sort of just a specification built on top of an underlying SDMX standard. So it's quite straightforward to get the two standards to speak to each other, which was sort of a serendipitous opportunity that we've been trying to take advantage of here. Now on the approach that we know this isn't the first effort to integrate survey data and estimates into HMIS, UNICEF, we've supported other efforts. I'm sure other partners in here have as well. And we've learned from those, but I want to just mention briefly how our approach is a bit different. One is the fact that this is bi-directional and mutually beneficial. So it facilitates the national level data use analysis and quality processes, while also expanding global databases with greater country representation. It's not the introduction of a new parallel tool. So what we're doing here is we're taking recommended triangulation methods and data workflows and trying to build them directly into the HMIS system or the platform that the Ministry of Health is already using. So it's not something where you learn a tool one time and then you're not using it. So you just kind of quit using it. We're trying to build it directly into the software and the platforms that you're already using every day. The country implementation in HMIS can also be customized. So we have custom dashboards, data processes, visualizations, and apps that you can develop from the back-end data. And it's been developed in partnership with the DHIS2 program, which as we all know here has a large expert community providing sustained support to countries for troubleshooting, customization, and implementation. And I think that community is prominently on display this week. So that's another huge benefit for this approach. Another thing that I just want to note is, since this is a proof of concept, and it's meant to inform a more long-term solution, we've taken an intentionally simplified approach. And by that, I mean there are some conscious trade-offs that have been made. So we've conceded that at the moment this will be limited to aggregate data and representative sample data. We're not going into individual level data at this time. And we're starting with a limited range of data sources, trying to start small. So this is not an interoperability layer or a massive data lake system. This is, you know, and so any country's interested in those other solutions, we would encourage you to still pursue those. We're just trying to start small, improve the technology and a large-scale sustainability and scalability of this. And so by conceding those, we've developed a solution that is extremely affordable, requires minimal training and capacity development, we believe, has low demands on existing ICT and data architecture, does not require the introduction of parallel tools, so it can be deployed directly in HMIS. And therefore, we believe this solution is highly deployable, highly adaptable, sustainable and scalable. So that's all the sort of theoretical concepts behind it. Now let's sort of get into what this actually looks like. So first, just sort of as a background, I want to just give a real quick snapshot of UNICEF's data architecture. So this is the current UNICEF global data system. It's called Helix. I don't want to go into too much detail, but do want to draw your attention to this middle box right here. Just to note that we have a data source catalog, which has raw micro data from thousands, probably tens of thousands of surveys and international models. And then that goes through a mixed data working system where we produce the indicator values and disaggregations. And then those values are pushed into a .stat SDMX warehouse. And from that warehouse, it can be pushed out to different various consumer interfaces. And I won't go into each of those, but you'll see that the blue one is the one that is relevant here where we've been working on specifying DHIS2 compatibility from our SDMX backend. And then just to go into the, yeah, so go into just a quick note on the mixed data working system as well. So this slide shows how the data get called out of our source catalog into an Azure Data Lake. We run an R package that standardizes and harmonizes the data across all of the tens of thousands of surveys. And then that harmonized standardized raw data set is maintained in a harmonized database also in Azure SQL. And then we have another R script that's run that can aggregate, calculate, validate, test, and visualize those indicator data and push them out to various interfaces, including the .stat warehouse. So you'll see all of this referred to simply as Helix, which is the name of our architecture. So that was the UNICEF databases. Now let's talk about DXP. So phase one, which was completed in March of this year, we developed the DXP, which is the intermediary platform that's developed in DHIS2. It's aligned to the core health indicators from the survey estimates and the core RMNCAH, WHO UNICEF package for in-country facility reported indicator definitions. And then we developed an ETL that can call data from the UNICEF global databases through the Helix has a RESTful API. And those data are restructured more or less to the ADX standard and then stored in this DXP platform where they can be made available to countries. And phase two is where we're looking to actually deliver our data to countries and support integration and triangulation in the national HMIS platforms. So obviously part of this is developing data sharing agreements and defining user groups and roles, but then mapping out and aligning the data elements and indicators and the organizational units. And we'll talk more about that in a bit. But then once we've done that, we can make our make our data available through the installation of configuration packages, which we'll talk about a little bit more in a bit. But basically there are configuration packages for the metadata and the dashboards, which are imported and mapped to the national system for the data elements. And then once that mapping has been done, you can import the configuration package for the data to populate the dashboards. And then those we're working on dashboard development and customization with countries. And then obviously as we're looking to implement this and user acceptance training and testing is important as well. And then once we've already sort of built this connection with countries who opt into this approach, we want to leverage these mechanisms to facilitate data flowing back from the country to the global level, to facilitate easier global reporting. You know, again, talking about data sharing agreements, user groups and roles, some additional alignment and mapping possibly, but then working out a data exchange mechanism based on the country's preference. And then of course, some training. So speaking quickly about the data exchange mechanisms. So between Helix and DXP, you already mentioned the development of the RESTful API and the ETL that calls data from Helix and pulls it into DXP. And from there, we can make it available to countries based on their preferences. So we have different options that you see here on the screen. We have the configuration packages, other ways to use the API, there's the data import app. And then similarly, when we're talking about pushing data from the country level back up to the DXP platform, where you can use the API again, data import app, or you can create standard reporting forms to do this. So there's a variety of mechanisms available to us. It just comes down to the country's comfort and preference and how they choose to and prefer to proceed on that. Again, don't want to go into too much detail here on the ETL, but just wanted to put this slide up here as well. So I think for the ETL, we're talking about how you in the script to extract it from Helix, to transform it into the proper structures and standards, and then load that into the DXP platform. So this has been developed in JavaScript using Node.js. And the documentation is here on our GitHub repo. So we have the link there. And so the vision is that we have this ETL, and we would probably end up developing another one to pull data back from DXP into the back-end data systems at UNICEF and other partners, like if we're working with other reporting partners that countries have to report to, where when we publish new data sets, if we publish a new data set every three months and update to our back-end data system, we can just publish that and push it out to countries so that they can have the most recent data available from our end back in their HMIS system. So briefly on the scope of indicators, as part of the planning, we mapped out the overlap between the global and national level indicators so that we could select priority indicators for this test case. And we found we have 45 core health indicators in our database. There were 48 core indicators in the RMNCH-WHO UNICEF package with an overlap of 30 where some triangulation, validation, or comparison is possible. But for the purpose of this concept, proof of concept, we've scaled down to focus on 34 indicators from our back-end databases and also 34 indicators from the HMIS packages with an overlap of 28. But I think it's also important to note here that in reality, we know that country tests will be focused on a much smaller number of indicators, maybe around 10 to 12 in most cases. And that's really due to the availability of data and the definition of indicators needing to be compatible in both systems. So there's a note on the indicators and data elements. We've also given consideration to how to align a sort of multi data system organizational hierarchy. So in green here, you see the UNICEF organizational hierarchy where we have global and then we have UNICEF regions and then we have countries and then we have subnational areas, usually down to admin one or admin two level. Whereas in a hypothetical country here, you have the national level at the top. Basically, your data typically goes from facilities to a district level and then often to either a region or directly to the national level. So the common areas here are the national level and the first administrative level. And so what we're doing is during implementation, we simply import the subnational organizational hierarchy from the national HMIS system. We're not trying to prescribe that for all countries globally, but if and when a country opts in to test this and implement this, that we would just import their subnational organizational hierarchy and then we can match the respective hierarchies at the national and admin one level. So that's it on the design overall. I have a couple quick notes here on what we've achieved so far and then I'll turn it over to John to start to walk through the demo. So where we are now is we've developed the DHIS2 platform that's aligned to the indicators and we've written the ETL to call data from the UNICEF back ends into DXP and make it available to countries. We've mapped out the indicators and aligned organizational units in our first test case and we've developed a generic dashboard app and user interface in the DXP DHIS2 platform, which you'll see here in a second. So our core results really are down there at the bottom. It's the HMIS compatible data structure definition in DXP and storing health data there, the customized data use app and successful remote staging and testing in with the Laos HMIS platform. Some other developments that we've had is just sort of, you know, we've had the opportunity to work with the development, the analytics and the quality teams at DHIS2 to look at how this approach can be adapted more broadly in DHIS2 systems, how this approach could potentially improve data quality processes in DHIS2, for example, basically creating a systematic mechanism for external validation and triangulation with more robust and independent data sources. And then also looking at sort of the analytics roadmap and what sort of improvements are already planned and what additional improvements could be made in the coming years to make an approach like this even better for countries. And then we've also, I want to draw attention to this note at the bottom here around data sharing. So this project has also led to some really interesting discussions at the global level with a lot of the major development partners and donors and multilaterals and bilaterals and these discussions on how to reduce some of the inefficiencies around some of the data sharing agreements so that we can reduce burden on countries while allowing greater data sharing with lower effort. So like I mentioned earlier, we know that oftentimes at different points throughout the year, you're submitting the same numbers to different agencies kind of over and over and you're just kind of having to do the same thing through inefficient processes and we're having discussions on our end to see if there are opportunities to leverage a mechanism like this and then for us to coordinate across the agencies at the global level to make that a bit more efficient and reduce some of the burden on countries. I won't go into too much detail here but just wanted to share the project timeline. Last year it was all about conceptual design, resource mobilization, initiating phase one which we finished earlier at the end of Q1 2020. We've also had as many of us have experienced has been a brief interruption of this work while responding to COVID but we're now moving forward again with our first country testing in Laos very soon so we hope which we expect to continue into 2021 and then we'll also be looking for an additional one to two countries for additional testing either later this year or early in 2021 so that we can ensure that the evidence that's generated from this approach is based on a sufficiently large sample of countries and and contexts. So with that I will stop for now and I will turn it over to my colleague John who's going to walk us through a demo of the platform and discuss this in a bit more detail. Thanks. John, I stopped sharing so it should be available to you now. Hi Al, if you can hear me now. Okay so I'm just going to try to show like what we have done with the DXP dashboard before going into the the dashboard itself I just wanted to go through a few of the things how it works I just have a bit of a background. Most of you are familiar with the DHS to data entry so this is the one of the screen like which we imported all the data from the survey so you can see all your the DXP survey data from the Helix system stored in your own local DHS to instance. So these are the few of the data items what we tried to get plus apart from that one you have your own country DHS to data. Like for example like for most of the places like it will be this is the from the survey data and then like you have the HMIs population and then your local country data elements and data sets which are basically collected at the district hospital level or the facility level and then once you aggregate that one then you are comparing your routine HMIs data as well as your survey data together. So to make that one happen what we tried to develop with the HIST Vietnam is a small app which is called integrated health data dashboard at the DXP which basically compares your HMIs reported data and your sensors and the survey data from the Helix system which is adjusted round across. So these are the few charts which we've been trying to to get through from the Helix system where you can also just see few of the potential outputs one is like comparing your key demographic data around and also you can also customize furthermore just to see your live birth data across different region. One of the other things why we didn't go for a creation of the using DHS to dashboard was we wanted to have the coding specific charts which is now available in 34-35 release but like before that one like it was not available but also like few of the other charts. For example like all the HMIs reported data are coded as blue and then the the sensors are in red and then the adjusted the one are in yellow. So this kind of things goes through across all the charts so it's easier to comprehend what data is coming from HMIs what data is coming from the survey like and how best we can practically compare across. So these are the few things based on the indicator what has been selected and the visualization. So we have been showing few of the charts and each and every widget can be downloaded into PNG which can be used for the presentation in the PPT. We are also making a focus on how best we can try to get the entire dashboard in a more PDF format or individual chart format altogether that's something which is going to come across. But further like more things like we are also included a few of the charts down here which is comparison of the session data post clinical care and this one is more about on the provincial wise coverage. So how is your HMIs data showing compared to your your survey data plus few of the the charts and maps. We also like since we are restricting ourselves to the national and the district level these are the few of the charts what we try to get with the different legends. So this is the basic overview and plus like once we have more and more data configured and the country needs we can include more charts type or the more team thematic areas which can be included around here. This is the basic demo like where we are trying to get the helix data and as well as show the comparison across the survey data and the HMIs looking data. There are a few currently like all the configuration is done in a JSON file where you map all the the data what is relevant and based on that when we are showing things up his Vietnam is trying to develop an admin tool so that like the the country admin person can configure the or map the data what is relevant from the survey data to the HMIs working data so that they can like have to get the similar kind of charts. Other things which you are also working on is on the translation which is also very key to try to to get it round for different both on the data side and as well as the UI to see every all these things in a multifunctional level using DHS to translation. Apart from that phone I guess like that's that's all from me so we can try to open it for the question. Oh sorry I'll hand over to Tyler again. Thanks John yeah please go ahead Tyler you can share. All right thanks so much John. Let me share my screen just really quickly as we wrap this up and then open it up for questions. I did just want to maybe just take one second first to just the as you could see in the dashboard that that John showed I think having those data side by side you know it's still this is our early prototypes and it's still a bit you know simple in the approach and we're continuing to look for ways to to strengthen and improve that but already you can see some of the value so there's several of those where you look at different different denominators some of the common denominators that we're using for many coverage estimates and you can see the variation by data source and then there are charts that flip those denominators just show what impact they might have on on coverage estimate so if you're looking at BCG depending on where you're pulling that live births estimate you might have 140 percent coverage or you might have 65 percent coverage and so having that understanding of what's in your system I think is really important to inform how you're interpreting and using the data in the system. There's also charts that show like by vaccine and by you know for different coverage interventions that the coverage estimate from the surveys versus the coverage estimate from the from the national HMIS which allows you to you know see how closely they align aligned they are and also to look at if they're become if they're converging over time to you know to indicate and you know improvement of data sources and also some would show you like even at the subnational level you know because we would expect more or less the same relationship between provinces when it comes to coverage estimate whether you're looking at data from the national HMIS or whether you're looking at from the national household survey program and in many cases we see that but you do see some outlier districts which you know you have significantly higher or lower coverage estimate depending on one of those data sources when they should be pretty similar so it allows you to have a little bit more nuance in interpreting the data and using that but also it I think it's a really useful tool for flagging data quality like potential data quality issues so that you can drill down into those a bit more and try to figure out you know if it's a if there's a there's one and they're like there was one thing in there that was like a 6500 coverage which it's blaringly glaringly obvious on the dashboard and it allows you to go in and say this is probably a data entry error and identify some of those challenges so thank you John for for walking through that and for all the HIST Vietnam support and developing and configuring that tool I also just wanted to acknowledge you know several people who have contributed to this so we've had UNICEF and WHO colleagues who have supported sort of the design and conceptualization of all this we've had generous support from USAID and the Bill and Melinda Gates Foundation you know obviously the whole team at DHS2 and the at UIO and the HIST group in Vietnam and then our colleagues from the Ministry of Health and Laos and the and the Laos office from WHO as well so just wanted to take a moment there to thank them for their contributions to this and then I think we can then go ahead and open it up for discussion thank you thanks a lot Tyler and thanks John I've been pushing a bit in the chat to post more questions we have two and I see one in the making okay we just got the third one in now Tyler I have one for you first from Carlos Herrera from the Ministry of Health in Honduras he's asking how can Ministry of Health Honduras do a similar implementation we have the HIS but next step is exactly this one integration and the exchange please let us know do you want me to do we want to answer each of these individually or do you want to just go ahead and other people please post your questions we still have more than 10 minutes yeah perfect yeah I know thanks so much it's a great question you know I think the first thing to do obviously is just expression of interest and we can connect I have my email here that you'll see on the screen now so you know for Honduras and anyone else who may be interested would love to start that discussion you know I there's there are some criteria that we typically look for in a country to make sure that we think it's a it's one that is a suitable test country but there are a lot of options that we have on how to sort of adapt this to deploy it in countries you know it'll depend a lot on the system you have in place the amount of data you have the amount of data we have for Honduras to try to make sure that it's sort of a mutually beneficial effort to implement this you know we you know basically we don't want to go through a ton of work and and start field testing this in a country and then we find out that oh this is like one of those countries we only have one survey for and so we're really not providing a ton of value or if we're giving you one 10 year old survey I don't think that's the case for Honduras but just sort of as a general note but yeah I think we'd love to have that discussion and look at the platforms you're using at the national level to see if it's one that we can that we can relatively easily you know work towards field testing and then sort start mapping out indicators and looking at the organizational unit alignment there are a few cases where the mapping of the the org units will be very challenging so like for example if if the DHS and mixed surveys that you're doing in your country are all based on administrative divisions but then you have some completely separate health divisions that are used in the Ministry of Health that's that's a bit more challenging it's not to say that it's not ever going to be possible but it's a lot of work for a field test for a proof of concept so that those may not be some of the most suitable options for for some of these initial one or two field tests but but most countries that's that's really not an issue and we can we can work towards finding a solution for field testing if there's interest in the ministry thanks thanks Tyler okay we have two more it's a bit more technical we have one from Joseph in Tanzania he's asking some countries allow integration of DHS to with other systems via an interoperability layer does DXP also adopt that that approach or is the integration don't appear to appear in example Tanzania is using health information mediator now like I guess that one of the things what we've been trying to focus on is how best you can try to compare the survey data with your routine hmi state with based on certain standards and the comparison for example the things which I'm just showing around here like in the survey data it shows it's 94 person coverage and the HMS is 79 more closer around here so in this particular province the survey and hms are nearly okay but this one is quite far where survey is 83 HMS is 19 so these kind of standard standard tools or comparison with the different surveys out there data coming from it's good to like have multiple source of data I guess that's also the what Tyler was addressing on the first question we can try to include many source but the first source we'll try focusing on is this one and based on that one we can try to use any other other source from and using the DHS to API you can try to always push the data into things we are not focusing on like you already know the DHS to standards where to push the data so we're just like making a a generic tool so that like you can decide how many survey or things you want try to configure based on this one that's something which is going to try to build on at the later part so there are supposed to be a bit more discussion we're testing it out with the with law with a few of the other surveys and the things so when we have one more pilot maybe in Tanzania in year 2021 like we can try to also see what all the other different surveys which you can try to be include and how the chart should represent when we are having three or four different sources of data to compare I have one thing to add on the ETL or the interoperability approach I think one one thing to important understanding is that we are not this tool is not an interoperability tool to use in countries between systems it's basically taking the global survey data in the Helix system globally and transferring it over to a global DHS to instance a repository and from there we can distribute this through the standard sort of import export approach in DHS through both for the metadata and for the data itself so the actual move to get this data into a country is much more the sort of the normal import export process with the DHS and linking it very similar to how we do the WHO packages being able to map it to data elements and importing data yeah I think that's a maybe you can say a little bit about how you did that in Laos John so people can understand yeah but based on the sorry I said Tyler was going to say something no go ahead John you go first yeah based on the the discussion what we had with the Tyler and with the Oula long time back so this was there they had the all the different indicators and also we wanted to to see like whether like Laos are collecting these data in the Rotin HMIs that was the first step and then we identified these are all the indicator which has been collected in the Rotin HMIs and we can compare that one to the the survey data and then we just say what are the additional surveys what we try to have so based on that one we try to to map the data so what can be compared between Rotin HMIs data and with the with the survey data which is which makes sense to to see side by side and see like how best the both of the systems are working around one of the the main things in many countries what we always just say survey data is always lower compared to the HMIs data HMIs data is always higher but in some of the indicators are nearly the same some of the indicators especially with the mortality and other things are very different but like it's just to see like how best we want try to to bring this gap closer the one of the way to bring this gap closer is by displaying this one in a in a dashboard way so that like not only the the only the key health administrators at the national level but also at the the program level and the things they can see like how things are happening across a different program at the program area. Yeah thanks thanks so much John and just to just to build on Ola and John's response like being able to to to combine data from your HMIs and your LMIs and your CHIS and your various you know other systems iris all those is really important and this is not the solution that's going to do that so that's why we continue to encourage countries to look at in-country and dropper ability layers that that can that can provide some of the solutions this is really meant to basically take a lot of global data put it in the same structure and standard that's compatible with DHIS systems and make a very simple process that can make that data available to cut to immediately you know 65-ish countries so that they can have access to that and do sort of a simple triangulation that that we've sort of demonstrated here but what I'll say too is if you're in a country like Tanzania where interoperabilities or layers are being discussed and considered and developed and so if you do have a system that pulls in data from your LMIs or your iris or you know other other information systems in the country there might also be additional indicators that we haven't included in this field test that would be relevant from our back end too so having that will also enable potentially a larger share of data from our data systems to the national level because it might allow us to you know work with our supply division and share some of our commodity data and then other logistics data that we could that we could sort of combine on but the sort of immediate test case is just looking at coverage data from HMIS compatible with our coverage data from our our global data systems. Yeah, thanks Tyler. I mean there was one more question a little bit related I think to what you just said. When from Zambia is asking in terms of the tool selection for the ETL tool why we use the custom Node.js script versus using some existing ETL tools out there. I don't think Stian is actually on the call he made that decision but I just want to emphasize it's a very very simple tool you can all go to get a look at it it's a very sort of basic process. I would also reflect on this a little bit say that what was really the complex thing here was to try to model the survey data on the DHIS data model because we design the DHIS data model for routine data and it doesn't fit perfectly well with the survey data especially with the time periods and being very different confidential intervals. A lot of this complexity is now sort of dealt with in this tool in the package in the metadata and the data you can download and I think that can make it a little bit easier for countries to use this data in sort of easy access way on the DHIS. A lot of countries have this data already in spreadsheets etc and have tried you know various levels of success to try to import this and use it in the DHIS before but I think that to me I think is one of the big wins here is that we can help try to model and integrate the data from a data modeling and design perspective maybe more than the ETL technical side. We don't have any more questions we have about five minutes. Tyler, John do you want to add anything? Yeah maybe if I could just build on that point you're just saying well let me find this I have some additional slides buried back here in the back and I want to talk about why we gave so much thought to so maybe first just to show this from the colleague from Honduras these are some of the criteria we look at for when we're talking about opportunities to pilot so you know does your country have DHIS2 or other ADX standard in use with the national scale have is there decent alignment with some of the global indicator definitions do we have ample data in our back end that we could provide and do we have subnational area alignment between sort of our helix structure and HMIS in terms of the data sources so that's where we get into like the administrative versus the health districts and things like that and then I wanted to share one more nope I guess I don't have it in here so there's a yeah that's unfortunate so basically what I wanted to show here I'll go back to I have a slide where this will work so basically the reason we put so much thought into how best to to model the data is because we wanted to have you can see it a little bit here where basically these these green indicators or these these green values here come from surveys and you see you have some that you know cover multiple years of fieldwork and collection and some that are in a single year but we also for each one you can you can see them here a little bit you have you know you're sort of median your point estimate with a lower and upper estimate for coverage from that indicator and we wanted to see the the way that the degree to which the the trends from the the national HMIS coverage estimates fit within these sort of expected bounds we would expect this this line here this blue line to cross through you know these these two data points within this sort of this box right here and so that required us to give quite a bit of consideration to how can you have like a single value that coverage is a range of time periods and has sort of a point estimate but also a lower and upper estimate and we want all of that to be something that we can compare to sort of a single trend line coming out of HMIS so that's as Ola said that's I think that's a big win is trying to figure out how to to model that and also how to be able to take data from an SDMX warehouse at the global level and make that available to countries because you know we're seeing it UNICEF we've done it the SDG database has done it in terms of moving to SDMX we're also having those discussions for the UN system as a whole I'm actually sort of working with the secretary general world's office right now on the implementation of a new data strategy which among other things looks at sort of the technology environment and trying to make sort of a common data back ends for all the agencies and departments and offices that can sort of have a more strategic asset value for the for the way they view data and make that available to countries and partners in the world so I think those are two of the big wins sort of related to the to the data modeling and sort of the the simple exchange mechanisms that have been developed right thanks Tyler I'm looking at this Europe we still have only those five I think we'll adjust everything I know many we are watching this as a recording on YouTube so please remember that you can still put your questions on the COP we'll keep the COP open of course a long-term tool for us to engage we'll also I think post updates there from the law experience and the idea would be if we can continue for those that are interested to follow this initiative and and post questions later or get updates to stay on the COP and and you know we can we can continue the discussion some communication there any last words Tyler or John before we close nothing for me other than just to say thanks everyone for attending and for your participation and you have my email there so if you have any other questions or want to discuss potential opportunities in your country or partners you're working with please don't hesitate to reach out and we'll be happy to have that discussion thank you great thanks a lot to Tyler and John for doing the presentation and thanks everyone for joining and asking questions and the next session we're starting six minutes thank you everyone and goodbye thanks