GitHub GitHub Universe 2020: Developer - Open Source Communities, 11 Dec 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: How the WHO uses technology for early detection, verification, and risk assessment of pandemics

Description

Presented by
Mala Kumar, Senior Program Manager, Tech for Social Good, GitHub
Bassem Dghaidi, Solution Architect, GitHub
Sara Hollis, Epidemiologist, World Health Organization
Dušan Milovanović, Lead Architect, World Health Organization

This panel will focus on the World Health Organization (WHO) and how technology can be used to improve public health. We'll discuss everything from Go.Data, a digital solution for contact tracing, to EIOS, a system that uses publicly available information to support early warning and response to public health threats. We'll also cover role of open source collaboration in WHO's early warning and response initiatives. Audience members will walk away with knowledge about how to engage with major public health organizations—including WHO, the Epidemic Intelligence from Open Sources (EIOS) community of practice, and the Global Outbreak Alert and Response Network (GOARN)—to help early detection, verification, and risk assessment of disease outbreaks to prevent pandemics.

For more from GitHub Universe 2020, visit https://githubuniverse.com

As always, feel free to leave us a comment below and don't forget to subscribe: http://bit.ly/subgithub

Thanks!

Connect with us.
Facebook: http://fb.com/github
Twitter: http://twitter.com/github
LinkedIn: http://linkedin.com/company/github

About GitHub
GitHub is the best place to share code with friends, co-workers, classmates, and complete strangers. Millions of people use GitHub to build amazing things together. For more info, go to http://github.com

A

Hi everyone thanks so much for joining our panel today, so my name is mala and I run the tech for social good program here at github and before I do anything else, I want to give a big shout out to my fellow social impact colleagues toya administ, cynthia campbell andrea. Thank you all so much for everything that you do now today. I'm joined by two brilliant colleagues from the world health organization dushan, is the lead architect of public health, intelligence and sarah is an epidemiologist with the health emergencies program.

A

Welcome to both of you and thank you so much for being here today.

A

I also want to give a shout out to my colleague bossam, who is a solution architect on our professional services team now bossam and I have been working side by side with who for the past couple of months, and he was meant to join us today as a panelist, unfortunately, due to circumstances beyond his control he's unable to be here today, and so we wish him and his family well, and we look forward to his return in january, if you're not familiar with tech for social good, we're a program at github that essentially ensures that our people, our products in our platform, are helping the social sector build the best possible tech solutions for some of the world's most pressing challenges.

A

You can learn more about us at socialimpact.github.com.

A

So to kick off the panel today, I want to talk a bit about an interconnected world in public health. If you're able to join us back in may for a github satellite, you might have caught the panel that I was able to host with some incredible social sector partners on covet 19 pandemic responses.

A

Now, during the presentation I talked about how we had connected with so many of our social sector partners to really understand how the pandemic was affecting them and their communities in the early days of the of the virus and from those conversations we kind of split it out into three major categories: societal, economic and pandemic, and while that's still really relevant to a lot of the work that the social sector is currently doing, it is important to note that today's panel focuses on public health and these categories are all closely interrelated.

A

So, if you're not familiar with public health as a discipline, essentially it's looking at the health and well-being of people in societal and community-based levels and a seminal person and a seminal event in public health was jon snow.

A

Now, speaking of england, jon snow was an english physician in the 1850s and in 1854 there was unfortunately, a cholera epidemic that broke out and killed about 800 people.

A

At the time it was thought that cholera was transmitted through little particles in the air, but jon snow had his suspicions, otherwise, and so what he did was he went and tracked each of the individual 800 people who died, whether it was in a hospital or whether it was in a school or a household or a business and in some cases, even a prison, and he plotted all of that on a map and from the work that he did.

A

He was pretty clearly able to demonstrate that cholera wasn't being transmitted through the air, but the contamination source was actually a well located on broad street in the soho neighborhood of london, and he was able to convince the local government that, because of this, they should take away the pump from the to the contaminated well, so that nobody could access that water source and almost immediately the cholera cases went down now.

A

This is important not just because he dispelled a really popular myth about cholera transmission at the time, but he also laid some really foundational work for a field called epidemiology which is going to tell us more about later is a field that looks at the the tracing of diseases like infectious diseases, non-infectious diseases and other health events like cancer in a population now john snow's work is seminal to public health and in fact it's still cited and used today now to state the obvious.

A

The world has changed a lot since the 1850s and in fact, if we look at data from the united nations department of economic and social affairs undeasa for short, which started reporting data in the 1950s, we can see, for example, some indicators that show us how how much the world has really changed. Now, one figure is population itself, so in 1950, when undesta started reporting data, we were at about 2.5 billion people worldwide fast forward to today, 70 years later, and that population is tripled globally. We're at about 7.8 million people.

A

Migration is another really interesting number back. In 1990 there was about 153 million people. We could consider asylum seekers or refugees or immigrants who went abroad for work or for school or for family.

A

Now, today, just 30 years later, that number has grown to 271 million, so that shows us the scale of which we're growing as an interconnected world. Now perhaps the most stark comparison is coronaviruses themselves right. We've all heard this word a lot over the last year or so, and coronaviruses essentially refers to a set of viruses that can cause anything from a common cold, all the way up to really severe respiratory diseases like covet 19..

A

So if we look at some of the coronaviruses and in recent memory and the diseases they cause, we get a stark picture. For example, in 2012 there was a coronavirus that originated from the middle east and caused the disease we now refer to as mers over about an eight year period, where there is the most active cases there are about 2500 cases reported and for my age or older, you definitely heard about a disease called sars, which came from a coronavirus in asia in 2003.

A

that caused about 8 400 cases in an eight-month period and for sure there's a lot of nuances with the numbers you see here on the screen in terms of the way that cases are counted so the methodologies that are used, the period over which you count them and then, of course, the actual viruses transmission and the diseases themselves.

A

But even if we take all of those nuances into an account and we we multiply the number of mers and sars cases by 10 or by 100 or by even 1 000. We are still not approaching the scale we see with coped 19.

A

as of when this deck was finalized, we're at 63 plus million cases and because we're in an active period of covet 19. That number is definitely going to rise in the coming weeks and coming months.

A

So all of this is to say that covid19 really reinforced the idea that the world is just too big. It's too complicated and it's too interconnected for any one country or organization to address public health on its own.

A

Now we're seeing really amazing collaboration in traditional public health, especially with covet 19 vaccine development and through our research and previous uh presentations here at github universe and satellite we've had the privilege of highlighting some incredible organizations that are really at the cutting edge of technology, public health collaboration, but even though that's been around for more than a decade, we know that, there's more that we can do to strengthen those solutions to further democratize those solutions and really to move to more evidence-based approaches.

A

Now we know open source is a really critical part of that journey and to that in github is really proud to be an implementation tool for public health technology, open source collaboration.

A

Well before I pass it on to the sean. I do want to point out that the world health organization lives on github at github.com. Backslash world health organization so take note of this link because we will be referencing repositories here throughout the presentation over to you, the channel, to talk about public health intelligence. I.

B

Need to say again thank you for a great introduction, actually making it easier now to jump into straight. What we're doing um again good morning afternoon evening, everyone um early detection events and their timely verification assessment of risks to provide actionable information for rapid and effective response to public health threats is the key to.

B

To mitigate and to even neutralize, if possible, um impact to society's economies, most importantly to save lives, and uh it's become obvious to everyone. I guess because the situation we're in with government.

B

This is the scope of public health intelligence discipline that I'm engaged with with my team in wga. So what kind? Next? What kind of events are we looking for? Well, um early detection verification risk assessment, as I said, is one of the strategic priorities of world health organization.

B

uh Current global situation we found ourselves in with kobe 19 has made it all the more visible and urgent and in response to ever-changing landscapes, including the way people communicate, relate interact.

B

Public health intelligence discipline adopted all hazards, one health approach. So what does that mean, for example, uh massive organization uh caused um that people connect with animals in nature much more than ever before? Maybe you didn't know, but 70 of known uh infectious diseases are considered, as zoonotic meaning that pathogens move from animals to people. Therefore, um we do not just detect and monitor uh for events that relate directly to human health. We also detect events related to animal.

A

B

Environmental health: um on the other hand, um we monitor not only disease occurrences in humans, but also hazards that have potential to cause health problems such as natural disasters, manmade disasters conflicts mass gatherings.

B

Next, um let me tell you, then, about more. What we're doing our starting point is data obviously, and coming from all kinds of sources, and there is no shortage of that. As we all know, in public health, we talk about different kinds of powerful surveillance. Information will come from different sources, such as healthcare facilities, missions, laboratories, pharmacies, news communities, field, workers, hotlines, some sources are publicly available. Some are not so epidemic. Intelligence from open sources is who led global initiatives.

B

The community of practice monitors publicly available information, mainly from news channels, but other public sources, and this goal is called public health intelligence from open source. So I guess you may ask, but why? Why do you monitor what you're trying to detect what is happening and events of input to help from all this enormous amount, mostly relevant information from all these publicly available sources like news, social networks, etc?

B

Why don't we use information directly from the source from the health care about the patients when they are diagnosed for precise, given conditions?

B

Well, data is actually not readily available from the source and when date is made available from through official channels, it's delayed to an extent that the delay itself can be published. So that's for early detection verification risk assessment.

B

That's why we are looking at different type of sources, publicly available sources of news trying to detect events as early as possible. Before I tell you what how does it look like? Really, let me tell you something about the history. So next.

B

Everything started when the web goes down soon after 1994 promed started providing emails to their subscribers with latest news related to public health going forward to 2009 a global health security initiative was a major milestone to organizing towards uh connecting content of these different public health intelligence systems that were developed prior to that in 2013 joint research center european commission, the same team who developed europe media monitor that you can see on the screen in.

A

B

uh Helped who to set up hazard's detection risk assessment system and that system uh evolved in 2017 into the system we have today. Its name is epidemic, intelligence for open sources, and that system uses not only information from publicly available sources, the world wide web and social media, but also information from these other public health intelligence systems. So, basically, in a way it brings together uh all information we have access to for the purpose of detection, verification risk assessment in order to inform and provide actual intelligence for time response.

B

So next, this is how your system looks like from the user's perspective.

B

On the left hand, side is a navigation panel, and there you can see a global surge of news, corresponding in time with uh kovite, 19 detection and confirmation um in china at the end of 2019, beginning of 2020.

B

article panel on the right hand, side contains a list of news- that's mathematically, categorized as in in near real time, within a natural language processing pipeline of eos system as containing content about hazards that that have potential for public health threats since call it 19 outbreak approximately 100 000 of these. Those kind of news are processed and stored in your system each day before we had influx of 5 15 000 news of that kind each day.

B

So you can see the research and and the amount of information we are dealing with since calling 19, eos searches for content within 30 000 news feeds and in twelve thousand twelve thousand web sources. Eighty languages meant for 240 twitter accounts and currently we have more than 40 million use items next.

B

um This is a sneak peek of how user sees eos system, so navigation panel left side actually provides text. Mining features for narrowing. The search for events that have been reported in use an article pan on the right side is made up of news filtered.

B

According to this criteria, selected on the left side and in this example, a user is selecting respiratory symptoms looking for pneumonia and unknown disease, and you can see that on a weekly basis from mid december 2019 to mid january 2020, the amount of noise is increased, is six volts and also you can see on the right side of the map that these news are mostly placed around this region of china next well.

B

Next, let me tell you well why we are here and what will tell you more at the end and why open source, but let me start with tell you that, whose main main objective actually is to help countries worldwide, as you know, to connect collaborate and share information and knowledge to improve health and well-being. So that's that's the solar role of who the key word. Customization objective will be strengthened by developing information infrastructure for connecting information knowledge in a global networking information.

B

So you know I told you how we detect today event. Potential threats is always is possible, and I told you that we don't have a possibility to connect directly source to to get patient health related data from health care systems and their individual durations made.

B

Obviously, if we would have had this kind of access and capabilities, then we will have even earlier detection of more relevant date. uh This issue is complex and uh it's important definitely complex and multivariate. Definitely uh we don't have time in this panel to talk about this, um but what can we do in the meantime? Well, in the meantime, one way is to we'll continue improving our detection from public health from publicly available sources, and we can mine volume and information using different analytical methods.

B

We can need a deck of.

A

B

Available publicly, however, we we have challenges we all this massive information includes conspiracy, theories, myths and disinformation. We have combination of opinions and facts. So how do we verify this information?

B

So we don't talk to me it's well, um alongside of analytics that we're using today natural language, processing, trying to discover content and context in uh this public related information.

B

uh We planned to go with linking data information knowledge available in different systems worldwide, such as those uh which provide public health intelligence and those which provide contextual information, and then we can, if we have this network information, uh semantically linked together, we can discover patterns in disease progression, spread, discover, similarities, unrivaled anomalies, while compiling to local data protection rules and regulations, and this is exactly a strategic objective of who going forward with public health intelligence.

B

So next, um how do we plan to achieve this objective of fostering connection and sharing data information? Knowledge?

B

Not only between countries and people as we're doing today, but between different information systems and data sets too well um we're working internally to establish a strategic which initiative to develop support of evolution of this information network interventions and this information network will power and be powered by the connected artificial, augmented energy systems to support not only detection risk assessment, but also to discover patterns of diseases and to learn to forecast in what breaks, maybe one day in future.

B

So by the end of this presentation, we'll tell you more about the type of open source projects that we have and give you more ideas about why these projects are seen as open source. But before that, I will um let you sarah tell you how the team uses technology to support fair workers in context tracing.

B

So with that sarah, over to you.

C

I'm to briefly walk you all through the go data implementation journey so far and how, in the context of the code event in pandemic, especially the rapid scale up of this tool now over 60 plus countries has really hinged on the collaboration of our user base through platforms like github and as duchonne highlighted as you'll see on the who github page once you visit.

C

This is just one of the many tools engaging the open source community in regards to pandemic preparedness and response, and hopefully it can provide you just a tangible snapshot into how these activities are starting to take place within the organization.

C

So, let's travel back in time for a moment to the pre-covet era, in particular in 2016, when the goal for this project really originated, there was a request from member state countries to who to develop a free lightweight tool for field level, outbreak, investigation and contact tracing.

C

This then began the iterative journey of requirements gathering and consolidating experiences from who and partner institutions that have participated in outbreak response over the years. The resulting platform in its more or less mature form was first piloted in 2019. During the ebola outbreak in north korea, it came here to an excellent piece that I really like it's created through esri story, maps platform, and it gives a nice overview of who's surveillance activities during this response just to hone in and focus on the cornerstone of this and any outbreak response. Next slide testing and contact tracing.

C

So once the confirmed case is identified, the contact tracing process begins and community level teams need to be quickly mobilized. In the case here ring vaccination of the newly approved vaccine was being introduced, which involves identifying and tracing not only first level contacts, but also secondary and tertiary level contacts.

C

So you can begin to imagine the value in having a really well functioning, well-structured tool for data collection that enables you to monitor hundreds, if not thousands, of contacts across the various aldusantes at any given time and how vital this could be actually in informing practical field-level activities on a day-to-day basis forward to 2020 when suddenly overnight, contact tracing is not just something you hear about in a presentation of ebola response, but I think that every country, country of all resource levels and across all geographies are talking about, and frankly, it was something that many countries needed to implement overnight.

C

Given the digital area we live in as well contact tracing apps became all the buzz, as countries were seeking innovative solutions to quickly face this crisis. I want to spend just a few moments to clarify what I mean by content tracing app for the purposes of this presentation. The term can be used pretty loosely in the media, and many of you in the audience's developers may very well be working on projects that fall into one of these categories, so who guidance on digital tools for contact tracing published back in june?

C

Lumped these tools into three main categories: outbreak response, proximity, tracking and symptom tracking. The go data tool falls into the first category and serves really as a reference outbreak response tool either for countries to use out of the box if they don't already have one or for countries to align their existing outbreak response to a width, thereby ensuring who standards on minimum data collection due to privacy etc.

C

So, in the interest of time, I'll really lie through a few features worth noting more information.

C

Demos are available online, but here's the initial web app interface and there are granular user roles and permissions that allow different types of surveillance staff to log in and interact with or access different types of data when they go to perform their necessary surveillance studies to outline quickly a few user personas a field epidemiologist may be performing case investigations through the web app and listing contacts which are then automatically assigned to the relevant contact tracer to perform follow-ups through their mobile and meantime lab personnel might be entering in lab results in a separate module, thereby prompting and alerting.

C

The surveillance team to then know which suspect cases are positive and which they need to perform a full-on case. Investigation for a supervisor at the national level can view an aggregate snapshot of the epi scenario. As you see in the screenshot here, or regional teams can dig deeper into the regional uh kpis.

C

The platform is built really on structured, integrated line lists so there's an example screenshot here it's really giving a centralized home to related case and contact data that can often be siloed across various systems, including lab and hospitalization data.

C

Next slide, perhaps one of the most important features is the ability to construct and visualize chains of transmission across the different entities in the platform through space and time so whether they be cases, contacts or exposure events. We also have geospatial mapping functionality, so you can see these linkages overlaid onto a base map of your choice.

C

The platform also illustrates the journey of cases and contacts longitudinally. So a few examples, screenshots here contact follow-up dashboards, the ability to see visually the sequence of events occurring for a given case or next screenshot honing in on key time intervals from a performance management perspective, so we're seeing here the time lapse between symptom onset and specimen collection.

C

So really, the point here is for the data to be transformed as much as possible into field level decision making and any operational improvements uh at any time that we can so um now across the past 11 months.

C

From our vantage point, at who we've been supporting governments and institutions globally to enhance their covenanting context, tracing activities through tools like godada and other activities, and this slide shows just a few examples of where I work with country teams on these activities, starting with drc last december, italy and early march, uh europe's largest refugee camp in lesbos in november and cantonbo in switzerland right here in my backyard.

C

So it's clear how vastly different these environments are, but the teams I've encountered along the way have remarkably similar challenges. Needless to say, there's a lot of power in sharing these learnings across settings and really not reinventing the wheel. In this way, community building around these activities has been an absolute necessity. We all know by now.

C

I think that the tried and true boots on the ground principles contact tracing, don't change all that much despite which digital tool is used and we've put a lot of effort in trying to convene a space for different users to interact whether it be through who consultations where countries present their strategies and challenges through our online community practice or through our github repository next slide. I think there's a screenshot of our community of practice and then yeah the next slide. We have our our github repository there, which feel free to take a screenshot of that.

C

So this is uh this repo, uh it's a central place where various users have begun to document and share technical resources developed and used in the context of their go data implementation, so individual users and institutions. We invite them to build on this library of resources and really adapt what they can for their specific setting and then share it back with the community.

C

So a few examples of what this entails. This can include various scripts and code snippets in r and python to produce automated analyses dashboards directly from the platform api. They have connectors to power, bi and other external analytics platforms.

C

We also have browser plugins for anonymization of data, middleware layers, kickstart, interoperability, efforts. Since we know it's, it's important that secure data exchange is needed in countries where more than one system might be taking place for routine surveillance and contact tracing, and we've really had the pleasure of working with a number of universities and organizations to begin to build out these resources.

C

So at that, we invite you to visit the the go data project within the github repo to see how you can take part. um I think that's it. On my end, I will turn it back to you mala thanks for your attention.

A

Great thanks so much sarah and duchamp for sharing some really interesting insights into your work. Now, as we've mentioned quite a few times in this presentation, open source is a really key part of the future of public health technology solutions and to that aim, the world health organization has been working really hard.

A

The past couple of months to open source, not just the work that sarah just mentioned, but also some applications that deshawn's going to talk about in just a few minutes now, there's obviously so many organizations, communities, people, individuals and companies that can plug into who work being such a critical portion to the public health ecosystem, but for the next couple of months these these are some of the categories of organizations that are really critical.

A

Now, of course, the journey to effective open source doesn't happen overnight, there's a lot of stages that have to happen in parallel and so uh just really quickly. Sarah, I would love to hear about your experience here where who is and how it's positively impacted your work.

C

Sure you know, um I think that, obviously I can speak for the whole organization, but dushan and I, as he alluded to earlier, and how I can certainly speak from my experience that we're both part of teams among several other teams at who that are very much early adopters of this type of collaboration within the organization. So this inner source kind of movement, that's been really cool to see.

C

There's been an increasing momentum around this so far, and not just across teams and departments and working across in ways that typically, they might have been siloed, but even through levels of the organization, so country office, regional office and headquarters, not not just with the emergencies program and the coveted response. But I used to work on a a team at who the who integrated data platform, which really streams inner source, it's integrated, multi-disciplinary team, doing a shared data platform shared reference data and standards across, I think seven departments.

C

So with that kind of stuff going on. It's really really a cause for optimism and it's really a cool thing to see.

A

Great thanks so much and then duchenne, I'm going to ask you the question that every it's on everybody's mind: how can the community get involved.

A

I think you're in me.

B

Excellent, I'm struggling with technology um yeah, so so inner source is one one really really important and theories and why we decided to to have a platform such as github for for fostering collaboration within who. But we have. As I mentioned, I mean during during my my discussion that uh um um understanding what's there uh in the world and what kind of information exists and bringing context together to understand, really what is happening and to react based on facts and requires really, as yuma mentioned, collaboration between different players and different people, and we can.

A

B

Going forward connecting the world and connect information data from different sources, we need those who have a grassroots approach, so uh one thing is: eminence-based organizations comes together, bringing things together, but on the other hand, if you want to react early, we also need to engage people and the more active we all are the better chance we have. So we have.

B

um You can engage with us we're working actively within who to uh to set up governance as we speak for open source collaboration before we set up everything we still open the repository that would serve as open source communication channel. You can see this on the screen and please take a screenshot or a link there. We can share um product roadmaps um uh collaboration guidelines and we can use issues of discussion to to engage early on with you. We already have a number of open source projects that you'll see on that. That page as well.

B

So there is, uh there is a scope for collaboration early on starting like 21.. Thank you to you.

A

Great thanks so much and just to reiterate dushan's point: please do take a screenshot of this slide so that you have the link. This repository is set up to really be your first interaction in case you want to get involved in who's, open source work. um I want to thank dusan and sarah for all of their hard work, these past few months to get this presentation together and really get all the work behind the presentation. It's been a long call and you all should be really proud of everything that we've accomplished.

A

um Also, please to our audience. Take a screenshot of the slide feel free to get in contact with us, sorry for the technical difficulties, but we really loved having you here. Thank you all. So much.