 All right. Hello. Hello. Hello. So good morning everybody. I know I have 12 minutes. I'll try and make this pretty quick I have a ton of slides. So bear with me I also before I start one of they say thanks to my co-authors of Pablo Marco Blanco Khaled Ahmed Cassana C's and Eric Fostermore. Oh and Tom Plaguey my brilliant software engineer and colleague So Everyone in this room probably knows what a difficult setting is and to give a bit of context as far as me in my background I started my career as a clinician went to school for stats and then epi, but There's this concept in the US this rising field of data science and this world of being a data scientist And it could mean you know a many splendid thing and in the context of humanitarian aid What a data scientist is and does has been fully utilized And when we think about these models of data collection and usage and consumption by this world We think on the academic side where something is is analyzed and retrospectively Sort of mulled over and published in that sort of world of the use of data and then the actionable Use of data for operations. So if we were to break data part into one operations and to academic and and Advocacy that's a good way to look at these next 10 minutes of my presentation on the sort on the operational side of things so As far as complete data solutions for operations and for action if we were to think about a program instead of data as a byproduct then you incorporate the use of data from the beginning from the design of a program whether it's nutrition or mental health or ncd Or an emergency project where you're just providing general care and aid in an outpatient setting to the collection stage where instead of you know a Lot of people in this room probably have a lot of field experience where you go into the field with an emergency project And about three months later someone calls an epi and the epi says oh, okay You have 50 excel spreadsheets all manipulated by a bunch of field docs and nurses and health workers Can you fix our data problem? And that's the norm It's not the exception. And so what we find a lot of times this byproduct of data usage Being a big problem so incorporating that into design and making data its own component of a project instead of instead of this output and then of course the storage of information so We suffer from it in the nonprofit world but also in the for-profit space that they've tried to really eliminate that the sending of identifiable information via email and In biomarkers and all those sorts of things and actually having a safer repository that meets, you know HIPAA standards and FDA and EU data security and then the analysis of information And and this is another big gap that we face in this context Which is a lot of people use you know tableau or click view or you know macros on an excel sheet for interpretation of information that needs to be digested in real-time and You have a lot of issues there from the manipulate the manipulation that can be subjective by excel sheet to the real-time nature of Analytics that's needed by field workers both online and offline and then dissemination by using something new Sort of shifting the way that we we digest information from one that you're expecting report on your desk If you're senior management to one where you can just log on to a website and see information coming from the ground in real-time So I'm just gonna scan past this but a lot of the issues that I faced both when I was at WHO and later at MSF was the overuse and over reliance on open-source tech Which which caused me to to rely then on consultants and and a lot of software engineers because I was awful at writing code And then a lot of the tools out there for for data management the humanitarian space focus on collection only Which forces people like me to make my own hack together storage and analytics components And then on the method side of the fence having This more complete nature of data usage from program design to analytics in a way That's built into the program was a big deficiency that I faced quite often both at WHO and MSF So the barriers I feel leveled Welcome to my life. So So anyone here that works in health informatics feels this pain viscerally where we have this big problem of is this Is any of this interoperable is one is one data set here interoperable than other probably not, right? So so this is a big this is the norm not the exception And then of course when we get into the field and from facilities into actually, you know new and protracted crises Both in camp and in the informal setting Collecting information here in a way that's statistically valid is a pretty difficult task and somewhat daunting Especially as the size of the crisis grows And then of course we have these these issues of not only collecting information at baseline about population needs But about the procurement and distribution of items that's needed and all of this information Ideally should flow to one singular repository where someone that's making the big decisions can make those decisions in a way That's informed by valid information And then of course there's the population so all the populations have their own Unique components as far as what the assessment looks like and how to get information both about their current and planned needs as well as barriers to care And when we think about what's needed on the technology side, and this is pretty important and to be a bit more Disturb the pot a little bit so I Am a firm believer in using the free market to our advantage as as Humanitarians so if tech doesn't work for you You don't need to customize it you can just buy the stuff that works And what I mean by that is if I were to look back in my last ten years of working in this field What I needed anyway was the ability to to manage my my data and not have to learn Python or JavaScript or something and And for that data to be managed from the time of creative program to the time I analyze it and give it to the people that have the big money to make the program decisions because I'm just an epi I don't make the big dollar decisions and then I need to have that user interface at field level both for the people collecting information And the people that are analyzing it at primary school level so analytics There's this big chasm between what a provider needs in the field as far as simple table charts and graphs To what is created in a lot of the for-profit spaces You use a tableau dashboard and you can manipulate it and it does all this cool stuff And then there's there's you know the high-level stat component and trust me I love multivariate models as much as the next person, but if I was a field clinician I just want to know how many patients I had that way how many died how many were deferred and what medications I need to give them and I need to be able to do that offline and then of course if I'm a manager in an office somewhere I need to create a project and deploy to the field collect my information view the results in a way that I know the information is stored In a way that meets all compliance and then I need to be able to to give that information to other people that I'm working with So what does that look like a hospitable data model would be nice Robust API both internal and external and then high-level aggregate metadata sharing versus via firehose So what we have a lot of times at MSF. That's a pretty big deficiency has been the the ability to use and utilize big data and That's not being utilized for the most part to be really basic because a lot of the systems in place at field level Are spitting up data that that sometimes is not valid or that cannot be Interoperable with other data sets across not only one operational center, but across the entirety of MSF and then of course, you know An opt-out data sharing provision, which is actually really important for some context So over the last four years. I've been working on a program called the HSP The original the original iteration was called the HAP which is the health health assessment project and then later on with MSF OCBA it was the health surveillance program and this was the first program that I know of its kind that that treated data as its own program just like ncd or mental health and The objective of the HSP was to provide rapid actionable information from crisis affected persons in the region So in the Middle East to better target current and future interventions. This was the objective originally by OCG so MSF Swiss and later by OCBA MSF Spain The pilot was in Becca Valley and 20 actually this is 2013 Sorry typo and it was in what we did before we scale this program as we we nested this this new technology and methodology in a larger traditional household survey and What we found using this new methodology was was a three times faster time to completion using a specific type of tech designed for this setting and There were no need of course if anyone's used mobile data collection tools Which I hope I don't have to convince anyone that paper is probably not the best You know way to collect and store and manage information if not, then I should probably go back in time a few years, but Mobile data collection tools are you know working in a lot of context and you know with over 90% of the world having smartphones at this time I'm gonna have to have to pull that card So you know smartphones are not just a developed world thing. It's a everywhere thing and we found This interoperability is the biggest takeaway, which is that we could do this assessment there and have the same methodologies same instrument same data set so a nice relational database in the back end and The users in Barcelona and Geneva could query the results regardless of setting and actually understand a bit more about the crisis in real time so we did this this baseline midline end line using a Google type user study and I have to scan through this part, but the presentation will be available to you later and Of course, we were able to get real time epitomologically valid information from very difficult settings A lot of places with no connectivity at all And then MSF was able to pivot faster and better to the needs of the population as it evolved and this is this is now at its third year second year with MSF Spain and Has been deployed in quite a few countries throughout the Middle East and this information is available real time via the web from the HSV dashboard So we looked at all different sorts of stuff from demographic information to health needs to barriers to care on vaccination coverage mortality morbidities pediatric illnesses vaccination coverage Food security livelihoods repatriation water supply and sanitation and in general migration and Then the way that this looked on the ground was serial cross-sectional assessments roughly a quite large sample size between six and eight thousand individuals nested in between 11 and 1200 households represented about the population of interest by the by the operational center and collected in trend so the actors can actually Assess whether or not the population evolution is is valid or not So mobile data question tools were used and let me actually skip ahead to what those look like so we used Tech it's called Dharma and you know full disclosure. I am at Dharma Because I believe in it and I am sick of using the over 200 Tools that I tried and tested while I was at these other organizations and frustrated with What Dharma allowed me to do was collect information wholly offline to track my staff offline to view results offline and to To actually manage my project Via something called a mesh network Then of course if there was one bar of connectivity everything went to a to a three layer encrypted database That was a for anyone the tech savvy multi-node postgres back ends a Jenga Back end that allowed me to Actually have these actors access a really simple web dashboard where they can view results both geographically and Across time and these results were displayed as simple graphs charts and tables and Available in real time if one bar of Wi-Fi or GSM was available on the web But again real-time offline on mobile real-time web if there was one bar of Wi-Fi or GSM they could also view their indicators and And and all indicators above threshold ping dread and track their staff so you could see where the staff were by hour by minute by day and most recent 6455 interviews in a five-day period many in North Babel And this was done a little little less than two months ago and throughout the Middle East so Iraq and Other sites I don't know how many I can name just for a variety of reasons But throughout the Middle East a bunch of countries think five countries and counting as well as now being in use by other organizations like the World Health Organization