[Audio Described] Using Data Science to Fight COVID-19 Session

– [Narrator] DSI Africa (light hearted bass music) Welcome to the DSI Africa State of Data Science Series We will be starting shortly Facilitated by Know Innovation Wednesday October 14th. In a video chat, a woman smiles Dumelang (Sotho greeting), Sanibonani (Zulu greeting), Bonjour (French greeting) – It’s wonderful this morning, and this afternoon for some of you, to host this webinar Welcome to the State of Data Science in Africa series which is presented by the National Institutes of Health Common Fund My name is Puleng Makhoalibe and I’m here with my colleagues from Knowinnovation We are also known as KI and like the KIStorm, the K-I in the KIStorm, our team based in the US and in Europe and also in Africa, facilitates workshops and meetings and online events, just like this one to accelerate scientific innovations And today we just want to start off by giving you a brief agenda for today, and it’s right behind me So first we’re going to start with some welcome remarks with a background on the funding opportunities And we’ll give you a few tips to make the most out of the experience on the webinar today on Zoom And we’ll give you an overview of today’s session on using Data Science for Fighting COVID-19 in Africa, from our moderator, Stefan Then a short presentation from each of the panelists that we have today, followed by a Q&A from our panelists And then we’ll have a plenary conclusion at the end But that’s not going to be all Immediately after the webinar we invite you to come to the interactive session that we host This will be optional for you but we do hope you will choose to join it And we’ll continue the COVID conversation with the panelists and network with others who are interested in the same topic, on how data science can help to fight against COVID-19 in Africa So today’s event is associated with the NIH program on Harnessing Data Science for Health Discovery and Innovation in Africa And I would like to hand over and invite onto the stage, Laura Povlich the Program Officer of National Institutes of Health at the Fogarty International Center to tell us more about funding opportunities and over to you, Laura – Thanks, Puleng. Just as a reminder and for anyone who’s new to our symposium, today’s event is associated with a new NIH Common Fund program on Harnessing Data Science for Health Discovery and Innovation in Africa or DS-I Africa for short There’re four different funding opportunities currently open for applications and you can learn more about these at bit.ly/nih-dsiafrica Applications are due starting in late November and we encourage you to visit the KIStorm platform to learn more about applying to NIH grants – [Narrator] Application receipt dates are November 24th to December 8th – Additionally, the contact information is listed on KIStorm and in each funding opportunity for each of the appropriate NIH scientific contact and we encourage you to reach out to us with any questions I’ll hand it back over to Puleng, to set up the rest of today’s session, thanks – Fantastic, thank you so much, Laura And if you do hear something interesting and you want to use social media to reach out to the masters who are interested in the same topic, please do use the three hashtags on this screen And those hashtags are #DSIAfrica and #NIHAfricaData and #knowinnovation So please remember those three hashtags as you go through two hours of this session with us today And before I hand over to our panelists and our wonderful moderator for today, I have a few things that I want to share with you So yes we are recording the session today and we’ll post the video on KIStorm which is the platform that you’ve used to get here and on the NIH YouTube channel

and in both cases captions will be added But if you have a technical challenge today, you can always scroll down to the KIStorm menu and we have a little page there that says “Need Help” and we’ll be there to sort it out immediately During the program we hope you’ll ask questions for the panelists during the session, and please add those questions by clicking on the Q&A button There’s speech clouds on the bottom of your screen and if you see a question that you think is a good one, you can let us know by up-voting it We are certainly a very large group today, I see we have over 67 people on the call And we would like to see, even though they’ll be numerous questions, we’ll do our best to answer the questions and as many questions as we can get to today Now, with that, I would like to introduce today’s panel moderator, Dr. Stefan Jaeger, Staff Scientist at the U.S National Library of Medicine and also at the National Institutes of Health So I hand over to Stefan, Dr. Stefan who is going to introduce the panelists and tell us more about today. Stefan, over to you – Oh, thanks for that introduction Puleng and it’s nice to be here today for this very interesting webinar You’ll see here, the title of the webinar is “Using Data Science to Fight COVID-19.” We will have five talks today that highlight how data science can help in the fight against COVID-19 and the methods being used, giving practical examples I will be the moderator introducing the speakers, and I will share my duties with Cecile Viboud, she is from Fogarty International Center, and she will also help be handling the question and answers, which will be at the end of the webinar I am a Staff Scientist at the National Library of Medicine At the NLM we are working on machine learning and biomedical image processing And we have actually worked on COVID-19 to detect that in radiographs like X-rays and CTs, and also we have a history in detecting malaria and other infectious diseases like TB So the topics for today are several, we will look into disease surveillance, measuring disease burdens, generating dashboards, text mining, data interoperability, and also publication tracking Also the NLM has actually a webpage, especially for COVID-19 publications, where you can find the latest ones but much work has been done on this too And at the end, we will have three breakout sessions with different topics on direct, indirect impacts of COVID-19 and also on the methods in data science being used to fight the disease (computer key clicks) This is the latest map of the World Health Organization regarding COVID-19 Africa is actually doing remarkably well, compared to other countries like Europe or the United States Actually South Africa has 50% of the infections in Africa and from Northern African countries as well But the rest of Africa actually is doing remarkably well, and maybe we can discuss at the end, why does this actually the case or data science can help find a wider scope, and Africa can for once be an inspiration for us, because it’s leading now in this (computer key clicks) Several reasons why Africa is doing better, I’ve listed here on this slide – – [Narrator] 1.5 million cases, and less than 40,000 deaths as of October 2020 South Africa accounted for more than 46 percent of the total number of casualties – could be the younger population, could be the existing infrastructure that’s already in place for other diseases It could be genetics, the lifestyle, the immune system, like the better contact tracing, we don’t know But I’m sure one of these or more of them are really the reason why Africa is doing so well (computer key clicks) These are the panelists today that we will have First, it will be Ali Mokdad from University of Washington, then Agnes Kiragga from Uganda Makerere University, then Elaine Nsoesie from Boston, Boston University School of Public Health

Then we will have Nirmal Ravi from E-Health Africa, and at the end we will have Frank Kagoro from, actually it says he has several affiliations, he is a research physician at Mahidol Oxford Tropical Company, Medical Research Unit and is also a PhD candidate in Cape Town, South Africa (computer key clicks) These will be our breakout sessions, – [Narrator] Direct impact of COVID-19 in Africa Is disease burden lower than expected and why? Indirect impacts of the COVID-19 pandemic, on other diseases and health resources Methodological advances to monitor and predict COVID-19 – as I mentioned, we’ll have one for direct impact of COVID-19 in Africa, one for the indirect impacts and then methodological advances to monitor predict COVID-19 You can actually choose one of these at the end or switch between them, we’ll see how it goes. And they should be very engaging, we will see And with this actually, I will stop sharing my screen and I will introduce our first speaker already Our first speaker is Ali Mokdad, he is a Professor of Health Metrics Sciences at the Institute for Health Metrics and Evaluation and he’s also a Chief Strategy Officer for Population Health at University of Washington As a public health researcher, Dr. Mokdad has published groundbreaking work on local-level disease trends and some of the leading risk factors for poor health and his work on obesity is among the most highly cited in the field As a Director of Middle Eastern Initiatives, Dr. Mokdad builds presence in the Middle East through new research projects, dissemination and uptake of methods and results and consultations with regional leaders in population health So with this, I actually leave the stage to Ali and looking forward to your talk – Good morning everybody, and thank you for that introduction – [Narrator] IHME COVID-19 projections for Africa: Strategies to reduce the spread – I want to talk briefly about IHME model for COVID-19 – [Narrator] COVID19.healthdata.org/projections – I will introduce briefly the model, drivers of the model, projection for Africa, some limitation and how countries can use these estimates I am from the Institute for Health Metrics and Evaluation We focus on what are the health problems of the world, how well society is addressing them and how could we maximize our output given the resources that we have The models that IHME started at the request of the University of Washington Medical Centers, they wanted to know how many beds they need, ICU, and ventilators, in order to prepare for the surge for COVID-19 patients coming here in my city, where we live in Seattle And we decided early on to follow mortality because of lack of testing early on And still in many countries they’re not doing enough testing We felt that test is a more robust metric And then we track to always model a peak and see what would happen after that peak how many ICU beds are needed And we also used early on mandates, later on, we have mobility, so I’ll show you some data And then how many infections we will have reported and estimated and we provide scenarios for health policy in each country – [Narrator] A graph shows reporting on a day to day basis in the UK On the weekend, the number of reports dip IHME has now implemented more than 3000 data fixes – We deal with a lot of problems in the data and here is an example from UK, England, the country that has a strong health information system and all of them historically have been one of the pioneers in that. But you could see the data here has problems, especially around the weekend where we have less reporting All what I’m presenting is available on our web and just Google “IHME COVID-19” and you’d get all this information, what I’m presenting today Am not going to go into the details about the models, I don’t have time for 10 minutes, but right now we have a hybrid model and then we’re using in our model to predict the future mortality and infection, mask use, cell phone mobility based on cell phone, pneumonia seasonality, which is strongly associated with COVID testing, population density, particulate matter, smoking attitude, and of course, pneumonia death rate And briefly we take the daily deaths and you see a lot of noises here on daily deaths this is cumulative deaths, daily cases reported, cumulative cases versus on cumulative deaths rates, hospitalization when we have the data and again you see

how noisy it is, and then we plot hospitalization rate against cumulative mortality rate And we make sense of all of this data, this is the redline that’s projected about the daily deaths, we smooth it and based on all this information together, we put this lead together to get a better estimate of daily deaths in each location One thing to be reminded that all of us, that in March, every country in the world, put some kind of mandates and the fact that they happened around the same time, it protected some countries because they didn’t have enough seeding So when New York shut down, it was too late in a way because there was so many infection When a place like Florida shut down and the same in Africa, there was less seeding and hence less infections So the mandate early on helped a lot for many countries This is in Africa right now What we see on the daily deaths – [Narrator] A graph shows a spike in deaths in the month of August, with a slow decline – You could see it’s around 3,000 right now, picked up sometimes towards the end of August, it’s coming down And here -R- Effective R in Africa, which is you could see it, it’s clearly, sorry, this is estimated population detected in every country Of course, South Africa have the most cases so far, so you see it has the highest number here Daily COVID mortality per 1 million on October 5, still it’s very low in Africa But the highest of course, as you’ve heard before it’s coming from South Africa This is mobility levels in Africa You could see that it’s coming back compared to January, almost back to normal, a little bit minus 10%, – [Narrator] Mobility levels as measured by smartphone app use compared to a January baseline – more so in South Africa There’s a pattern here in the southern part of the continent, you see more less mobility and as you will see later, more mask wearing This is how many people wear a mask and many places it’s like they’re 30% and in South Africa and the southern hemisphere of Africa, it’s almost over 70% Tests here is very low in Africa, as we know It’s high in some countries, but usually it’s very low, per 100,000 people, so not much testing going on for that early detection of COVID-19 And we present for each country and of course for the whole continent, which I’m showing you right now is for Africa And Africa in my definition here is WHO Afro region So this, the first one is mandate easing let everything would be normal and then we keep opening for business Our current projections which is our preferred scenario is that we would go into a lockdown when deaths reaches eight deaths per day per million per population And universal masks, if 95% of people wear their mask, what will happen in the country for COVID-19 estimated move down So for Africa between now and now we’re projecting to February 1st, what we’re projecting about 48,700 deaths in Africa, again, Afro Region – WHO If we keep easing the mandates about 49,000, not much difference here, because Africa already has low rate and it’s declining And if everybody in Africa, 95% wear a mask, we estimate that drops to 32,600 deaths So a huge difference and a lot of deaths could be prevented, simply by wearing a mask. And as you have seen in Africa, it’s still very low The limitation that we have, it’s under testing and under reporting of mortality of course we know there are problems, vital registration, in many countries that I’m showing And of course we don’t have, in our estimation of bed capacity We don’t account for surge unless a country provides us an updated number about the surge in ICU and their bed capacity We don’t know from the testing, how the sample is done, who is being tested, is it people come to a hospital, is it people who’s contact tracing So we have no idea about the dates, who’s been tested And in many countries, unfortunately we don’t have data, especially daily data about hospitalization And there are not many surveys in Africa, few of them about sero-positive, how many anti-bodies have been found basically to tell us how many people are infected So in multiple location and including Africa, I’m showing you here Delhi and Pakistan, we have seen that the pandemic peak and coming down – [Narrator] 2 line graphs show simultaneous peaks in infections in both Delhi and Pakistan – And we don’t see any association between

that lack of decline and mandate So you know a mandate happened early on, as I showed you, then the peak came later on and now it’s coming down And Stefan provided a slide, so there’re several explanation, I’m adding one here, role of super-spreaders in any country that have been saturated and they’re not any more actively passing the virus to others But biggest theory right now that people are zooming in on is cross-reactivity to other coronavirus infections So this I will remember when you see COVID-19, therefore there is less cases or less symptomatic cases in this case and less deaths of course, of COVID-19 Of course we have to anticipate a large-scale indirect impacts of COVID-19, domestic violence, women’s health, child immunization, chronic condition, of course on education and unemployment and we’re afraid, of course, all of us on gender equality, of course, and other type of inequality Immunization, I’m giving you a study from MMWR here in the U.S Even in the U.S. we have seen a rapid decline in immunization, unfortunately, – [Narrator] A bar graph shows non-COVID vaccine administration dropping in March – after COVID-19. Starting in March here, you could see it’s like clearly that the immunization dropped down compared to the previous months Under reporting excess mortality we see it in every country, the red here is actual mortality the other line is showing COVID-19 deaths You could see there’s a gap and there increasing, excess mortality increases as COVID-19 mortality increases Recommendation, how people use the data for planning, of course, because we provide the hospital, but also mandates, mask with penalty, that’s what is working in many places, also do sero-positive surveillance so we can track how many people have been infected, report what’s happening in the hospital And include the surge capacity so we can add it to our forecasts and for every country to be able to know how much your bed capacity is that’s the key issue here You don’t want to overwhelm your medical system and ICU capacity should be left at least, don’t go above 80% otherwise there’ll be problems, because we need these ICUs for other conditions And then don’t be afraid to impose a mandate, and then we can model other scenarios for any country Thank you everybody – [Narrator] Email, MOKDAA@UW.EDU. Twitter @AliHMokdad – Thank you, Ali, great nice talk, indeed A quick reminder for everybody, we will collect all the questions to the end of the webinar, and I answer them all in one session If you have a specific question you can enter them in the question and answer box, if you look at the bottom of your Zoom window, there’s a chat box and there’s also a Q&A box So please enter these questions in the Q&A box, and we will answer them at the end of the webinar, before the breakout sessions, thank you So the next speaker will be Agnes Kiragga I will shortly read her short bio here So Dr. Agnes Kiragga is the Senior Research Scientist and the Head of Statistics at the Infectious Diseases Institute at the College of Health Sciences, Makerere University in Uganda She was nominated as one of the prestigious Next Einstein Fellows, for the year 2019-2020 and featured among the top 20 African Innovators by Quartz Africa Dr. Kiragga spearheads HIV research projects including the use of routine clinical and administrative databases for prediction of outcomes among persons living with HIV as well as prevention of HIV acquisition Her research involves the use of data science methods to address various health challenges in Africa by leveraging their resources at the NIH funded African Center of Excellence in Bioinformatics and Data Science Dr. Kiragga supports initiatives for increasing the participation of African women in science, technology, engineering and mathematics, STEM So with these words, Dr Kiragga the stage is yours We are looking forward to your presentation – Thank you, Stefan I would like to thank the organizers for giving me the opportunity to talk about our experiences in mining social media data, and as well as understanding the indirect effects of COVID-19 on other diseases in Uganda Yes, so like the previous speakers have mentioned, Africa still suffers from a couple of COVID cases In Uganda the first COVID case was identified in late March and to date we have hit about 10,000 COVID cases that have been identified

in Uganda and about 95 deaths so far The recovery rate is about 63% with very low case fatality rate and about 12 people per 1,000 people are tested for COVID in the country So like any other country, there’re several non-pharmaceutical measures that have been used to control the spread of COVID There have been several measures and right now the country is embarking on measures to combat COVID We’ve seen various drives by telecom companies, making out deliberate efforts to ensure that the community ensures face masking and most importantly physical distancing as these are important for epidemic control What we’ve not really seen is how the community responds to some of these measures and how we can use it as science methods to understand what’s happening in the perspective of the community regarding COVID control in the country And so what we decided to do was to look at the social media platforms, and I got this slide that shows that between October 2019 and September 2020 there’s been an influx or an increase in the rate of use of certain media platforms Facebook remains the top media platform that’s used in Uganda. However we see in around April, just after the first case of COVID was identified, there’s been a spike in the use of platforms like Twitter And we hoped to leverage this data to see what is happening in the context of the Ugandan community regarding COVID Mining of social media data has been done elsewhere, and what we did was to leverage the Natural Language and Processes techniques to process Twitter information on COVID-19 We picked out tweets of conversations that were available between March 1st to date And because the activity started quite late we then went back and purchased a historical data of Twitter between March and August 27th and I’ve continued to stream this data moving forward With particularly looking at tweets that are geotagged to the coordinates around Uganda and the hashtags that we’ve used involve #COVID19, #STAYSAFEUganda, #COVID-UG, and several other hashtags We have collected about 100,000 COVID tweets and perform sentiment analysis to understand the positive and negative information around COVID-19 in the country And what is interesting is that when we looked at this data, we see that most of the tweeting happens over the weekend as expected, people are busy during the week, and this could be a weekend effect because some of the most common words that we see are affiliated to religion, or church, or prophets, going to church, but most importantly COVID-19 as expected becomes one of the most common hashtag that we’ve noticed in the time We’ve also seen that as the country goes into political elections, along the way, people start to talk more about elections And then this slide tells us ideally, how the sentiments of people have been evolving over time in relation to the COVID cases So the red line on the left box, the red line shows the COVID cases increasing exponentially and on the blue line, that is sloping from the left to the right then shows the sentiments of people So as the cases increase the Ugandan community then the sentiments reduce, so we kind of get negative sentiments around COVID as expected there’s fear, there’s anxiety And the same applies when we look at death With increasing death, the sentiments change over time People get anxious, people are fearing, and it’s important for public health people to see what happens in the community And then what we also saw is that there are several influencers in the Ugandan Twitter platform The most common of course, is the Ministry of Health, which is the top left highest bar on the left on the bar graphs Ministry of Health has consistently shared data and updates on Twitter And because of this, it’s one of the most influencers on the Twitter platform And then there’ve been several other influencers that are very known. Persons in our the community, for example, the second person is a well known inspector-general of police and who eventually got COVID So just understanding his sentiments along the way when he was going through this COVID disease while it’s important. And we also see that with time most of his sentiments centered around being trustful of the system And if you have influencers like that, can be very influential in informing community,

how to respond to COVID And there’s much more that we can do with this data, but for now, I’ll just stop here But what is important to know is that mining those Twitter has a great potential to inform public health response to COVID, the community perspective using sentiments is very informative, as healthcare planners and the Ministry of Health in Uganda can then leverage these sentiments or influence us to then inform COVID control Religious heads and people who are in the spiritual realm can then be used to promote non-pharmaceutical measures for controlling COVID At a time when churches are being closed for long, it’s important to know that we can then use some of these media platforms to inform the public This data is of course has some biases because it’s incomplete for the stat, it’s based on geolocation, which might miss out tweets in case the users do not have their location turned on, and because of this, there’s ongoing work to increase the pipelines and to improve the pipelines for streaming data to improve its completeness And again, we want to look at the COVID-19 myths in the Ugandan context, and also look at other social media platforms So I’ll then move to the second part of the presentation, which then looks at what has been the situation of COVID-19 and HIV in Uganda This work is from my recent publication, where we predicted the impact of COVID-19 and the potential impact of the Public Health response on the disease burden in Uganda. That is published in America in Journal of Tropical Medicine, please check it out if you get some time So, as a brief background, again, like the previous speakers have been saying, the African continent has experienced fewer COVID cases compared to what was earlier predicted and compared to other regions and other continents Africa has a 1.2 million, and again, there are much lower deaths, for example, compared to the U.S and like, we think that because of the previous existing infectious diseases and fragile health systems, we thought that the impact of COVID would be much worse. But what is important to know is that there’s still limited data on the impact of COVID on other health diseases and looking at the burden of COVID in Africa So to do this, we then pulled out data that was available at that time, and we leveraged the existing datasets of COVID cases, and most importantly data on important indicators like the HIV reporting data But what is important to note is that when you are estimating the burden, we then use DALYs, which are the Disability Adjusted Life Years Because these are very useful when you’re estimating disease burden, because the age structure of African continent is much lower, which is about 50% It was very important for us to relate this to the disease burden And particularly looking at the DALYs lost, if a person did not necessarily get COVID, but then suffered from any other illnesses that were as a result or increased as a result of the response towards COVID We use the WHO global burden of disease estimates and then looked at the excess HIV that would- particularly the HIV cases that would have occurred in the situation of COVID And most important, look at what happened during the lockdown in Uganda and we use data that was available at the time when the lock down had just occurred And just to give you a brief of what we think was happening, this graph looks at the data from 2019 all the way up to April 2020, when the country had just locked down And we see a very steep dip around December, which is expected, that’s the first dip But the second dip is at the time when the country has locked down We see that there was steep reductions in the number of newly identified or diagnosed HIV positive persons and steep drops in the number of patients who initiated ART at that time So we use the data as of that time to understand the pandemic And what is important is that the DALYs that we think that because of the data that was used at the time, the country loses about 405 DALYs due to HIV And we also saw that HIV morbidity and HIV mortality increased a notch about 465 DALYs lost And this is a big impact in the country, and we think that this situation has changed over time because that’s the recent data that we have right now But in conclusion, we think that looking

at the health service is delivery in terms of the COVID response is important And looking at the long term and short term impacts will be critical in understanding the response I acknowledge my institution and the NIH and the African Center of Disease Section. Thank you – Thank you very much, Agnes For the sake of time, we will proceed immediately to the next speaker Again, please you can enter your questions in the Q&A box So our next speaker will be Elaine Nsoesie She is an Assistant Professor of Global Health at Boston University School of Public Health She has a PhD in computational epidemiology, a masters in statistics and bachelor in mathematics Her research is focused on the use of digital data and technology to improve health and communities globally Her work has also focused on addressing bias in digital data and understanding factors influencing disparities in health outcomes She is on the advisory boards of Data Science Africa and Data Science Nigeria She’s also the founder of Rethe’, if I pronounced this correctly, an initiative focused on providing scientific writing tools and resources to student communities in Africa in order to increase representation in scientific publications Elaine, please go ahead with your slides, thank you – Thank you Stefan and thank you Cecile for inviting me to be part of this So, as Stefan mentioned, I’m going to talk about some of the work we’ve been doing on using such data to model COVID in South Africa So most of my work focuses on digital epidemiology and basically what that means is that I look at how we can use data from nontraditional sources, this includes internet sources, remote sensing for public health purposes, specifically civilians And so I want to know how we can use what people are putting on digital platforms, to study the patterns of disease, injury, and health and drivers of these patterns in specific populations (computer key clicks) So I work with lots of different data streams, these include text images, video, biological data, and this data comes from a range of sources, search engines, social media, being some of the mobile applications, crowdsourcing, et cetera I’m going to focus specifically on search data for this talk, and there’s so many different uses for search data, especially in the African context So some of the earlier work that we did during the pandemic was looking at what people were searching for So what kinds of information were people lacking across the continent? So we looked at 21 English speaking countries to see has the information that people were seeking changed over time. And so when you had at the beginning when there was a lot of conversation around the need for testing, there were people asking questions about how to get tested, where to get information for testing, do they have to pay for their own test? So those are just examples of ways that we can use this data For this particular talk, I’m focusing on how we can use it for disease modeling So a brief background on this data in South Africa As of October 12th, South Africa had reported about 693,000 COVID cases They’ve had more than 4 million tests conducted and over 624,000 recoveries (computer key clicks) – [Narrator] “The adoption of these digital platforms for public health surveillance creates complex challenges, requiring the development of solutions specific to the sub-Saharan African context.” Nsoesie, Digital platforms and non-communicable diseases in sub-Saharan sub-Saharan Africa. Lancet Digital health. 2020 – And using digital data in this context means we have to think about the specific cultural context in South Africa. And we can’t just assume that what we’ve done in the U.S. for example, will work in South Africa So we have to think about how do we bring in the cultural influences in South Africa to include that into a model so we can build better models for estimating the spread of infectious diseases in this case, COVID So we started off with general questions So these are things that you see anywhere, not just in South Africa So all kinds of questions that people asking about COVID so things like what are the symptoms of COVID, symptoms of flu. But also specific symptoms that have been noted for COVID – so things like cough, headache, shortness of breath, fatigue, vomiting. We looked at that data And then we also looked at other ways in which people searched for COVID – so coronavirus, coro, how, things like COVID-19, also specifically And then home remedies, which I think this is where context really comes in

So we sourced these from colleagues in South Africa So we asked them what have people been talking about in terms of home treatments. And ginger tea is an example that kept coming up Eucalyptus oil is another one And this changes over time So there’re times when people would talk a lot more about ginger tea and then other times they’ll talk more about eucalyptus oil And this is all very important for the models that we’re trying to develop – [Narrator] A chart shows the prevalence of specific search terms from province to province – When you look at the provinces, so we have nine provinces in South Africa, there are very significant differences in the different provinces So yeah, we have two provinces and I’m going to present examples for these two provinces only since we don’t have time to go into all nine provinces And then on one side you have all the terms, all the phrases that we looked at So zinc is one that is very highly correlated with COVID trends. So yeah, you have the correlations on one side going from around negative 0.5 to almost close to one so 0.9 in some cases So you have zinc being very highly correlated You have things like loss of smell being highly correlated with COVID cases, so there’s a such use for those terms You have searches with COVID symptoms being consistently correlated across most of the provinces And so, because there’s this very significant differences across the provinces, it was important for us to fit models with each of the provinces (computer clicks) When you look across the nine provinces, there’re search terms that keep popping up, so things like COVID symptoms, shortness of breath, eucalyptus oil, are repeated multiple times, zinc, loss of smell, runny nose So these are terms that seem to be a highly correlated with COVID trends, so these are things that people are searching for over time We used the LASSO model – [Narrator] LASSO is least absolute shrinkage and selection operator 2 mathematical formulas are shown – So our interest is doing very simple regression, and so yeah, our Y, T represents COVID cases X, I, T represents Google searches so we have multiple search terms and we only want the ones that are most significant So our model in this case is doing two things One it’s allowing us to select the most significant terms and then it’s also allowing us to fit a regression model with multiple variables (computer clicks) So the terms that are not significant in the model, basically have coefficients starting zero and they basically get dropped out of the models So this again is the first province that I showed you before And results were significantly different across the different provinces, as you see for the two examples that I’m showing you – [Narrator] Results varied across 9 provinces 2 graphs show results from Gauteng – So on the X axis we have proportion of positive cases, so instead of focusing on the raw cases, which we did look at, we wanted to look at proportional positive cases because testing changed over time So we wanted to just look at how many cases were actually being reported as positive from all the cases that were being tested And then we have our model estimates on the Y axis, and then you have our confidence intervals around our estimates You can see that a lot of the points that we’re estimating do fall within our confidence range And then on the right side, you have two lines So one is the model and the other is the cases And so the long black line that you see that indicates where we stopped the model feed So we feed the model using data from when we had the initial case reported in South Africa, up to that time point where you have the black line And then right after that, we started doing one step ahead predictions One thing that I think is interesting to note is that a model peak happens before the cases And one possible reason for this could be that people were seeing symptoms and searching for that information online, but then they don’t go to get tested until later (computer clicks) – [Narrator] Graphs show results from Mpumalanga – So this is the other province You can compare the two Again, on one side we have model estimates and proportion of positive cases and confidence intervals around those estimates On the other side, we have the trends over time So we have the model and we have the cases, you can see this as actually very different in terms of predictions after the model has been fitted Also, we don’t really see that delay in the peak, so it’s not very consistent from province to province, which is why we’ll feed in different models for different provinces And for some provinces we have very good fit

in terms of, if you look at how the model attracts the cases, while in other provinces, that’s not the case, especially where we start predicting So these are things that we’re investigating a bit further So some of the reasons why you can see differences in provinces include things like internet access, so how many people actually have access to Google, how many people have the time to spend looking for things online Demographics could also be a factor and also COVID burden, so provinces where you have a high burden people might be paying more attention than in provinces, where you have a lower burden Another thing that we’re trying to do is do a real time assessment of the models that we’ve developed so that we can have better models for each of the provinces and also just see how useful this data is in South Africa and all the countries in Africa I want to acknowledge all the contributors in this project, especially Brooke Nichols who is in this project and doing amazing work, supporting COVID response in South Africa So thank you all for listening – [Narrator] Email, ONElaine@BU.EDU – Thank you very much Elaine, fantastic talk Again to all the audience, please, if you have questions use the Q&A box at the bottom of your Zoom window We proceed now right away to the next speaker, Dr. Nirmal Ravi Dr. Ravi has a background in biomedical engineering and primary care Dr. Ravi started his professional career as a biomedical engineer designing wireless ECT monitoring systems for resourceful settings He went on to do a PhD in tissue engineering and MD from the University of Kentucky He has lived in several African countries over the past 13 years working on several diseases Dr. Ravi joined eHealth Africa in 2016 where he’s currently the Director for Medical and Scientific Affairs and Chief Innovation Officer for their clinics He leads a portfolio that includes laboratory diagnostics, electronic health records, clinical research, technology, adoption, and community health program Okay, with this please Ravi, go ahead with your talk – [Narrator] eHealth Africa COVID-19 Response – Thank you, Stefan, for the kind introduction and thanks NIH for giving us the opportunity to present So, I will talk about our response to COVID-19 in Nigeria eHealth Africa is a non-profit that has been working in Nigeria for the past 10 years and then the last two years we have also started delivering direct healthcare through a subsidiary called EHA Clinics So I will talk about the different products and services that eHealth Africa is offering to assist the Nigeria CDC in monitoring and responding to the COVID-19 outbreak These are some of the services that I will touch on We have a logistics management product that Nigeria CDC is using to track commodities for the response We have developed an automated symptom screening tool We’re assisting in interoperability of lab testing data through a lab information management system And finally we are assisting in sample collection as well as PCR testing for COVID-19 So, I will start with the data and logistics support that we provide The first tool that I will demonstrate is an interactive voice response system for symptom screening This is an automated screening for quarantined individuals which is especially useful currently in Nigeria as the country opens up to international passengers So the mandate is that every passenger that’s coming from abroad to Nigeria and a lot of them are Nigerians that are living abroad coming back home, would need a negative PCR test to board a flight But after arriving in Nigeria between seven and 14 days, they need a negative- they need a PCR test results And between those 14 days they need a daily follow up as part of quarantine The tool allows the user for bulk import of persons-of-interest records So, just from the flight manifest the tool can import in passenger data and then can automatically follow up The good thing is- or the advantage is- it replaces the manual process which you can imagine can be very time intensive, following up on thousands of passengers, hundreds of them every day would be a manually intensive process, because the past majority of them would be asymptomatic So what the tool allows you is to focus human resource just on persons who need additional attention because they screen positive on the symptoms There’s a dashboard that allows an overview

of the statistics as well as any real-time alerts that can go to surveillance officers The system is interoperable with a national digital surveillance tool which is SORMAS Alerts can be sent from the same time screening tool directly to SORMARS which can then be followed up with surveillance or by surveillance officers The system can be easily modified for any workflow and the responses and questions can be recorded in any language We use an off-the-shelf tool to develop this particular screening tool so it’s customizable to the user needs I will demonstrate the tool here – [Narrator] He clicks on an image of a smartphone [music] – [Narrator] The phone receives a call. He accepts – Hello, this is an automated call from the COVID-19 Response Team at the Nigerian Center for Disease Control We are following up on your health status I’ll be asking a few questions and it is important to respond Please verify that this is you, Douglas Moran If this is you press one for yes, two for no, [beep] – [Narrator] He presses one – or three if this is a wrong number This is Day 8 of 14 Do you have a fever? Press one for yes and two for no [beep] – [Narrator] He presses two – Do you have a cough? Press- – So, that gives you an idea of how the tool works Any person who answers positive for one of the screening questions is then tagged and would be visible on the dashboard as you see on the screen So this gives you an overview every day how many people have been called How many are responding How many dropped calls How many failed calls And then how many people will need follow up So far the system has made about 69,000 calls for about 10,000 arrivals out of which about 1,500 people screened positive on the questionnaire On the dashboard you can also see on a daily basis how many patients are screening positive on the calls and these can then be passed on to surveillance officers for follow up The next tool that I will demonstrate or briefly talk about is the commodity tracking tool called LoMIS and as Logistics Management and Information System For a large response like COVID, you can imagine there are millions of items that are test commodities, personal protective equipment that are coming in at a federal level and then are distributed to the states Currently, the tracking for this is manual with reports being generated manually and sent to the central monitoring station We have provided an automated system again that keeps track of inventory, real-time stock levels, utilizations, and expirations, so things that are first-in are first-out to prevent expiration of items on hold We have recorded about 1.6 million items so far with this tracking system Again, this has a dashboard that shows you how many items are received for each of the different categories, how many are outgoing and how much stock is being held – [Narrator] Total Incoming Commodities, over 950,000 Total Outgoing, over 460,000. Total balance, over 488,000 – I will now move on to the actual sample collection as well as testing that is done by EHA Clinics in two locations in Nigeria And these are our sample collection and lab From our test data we have tested about 12,000 samples The data on screen shows only about 8,000 samples To put this in context, Nigeria so far has tested about half a million samples in comparison with South Africa, that previously one of our presenters showed about 4.4 million So much lower testing rate The test positivity rate is about 10% About 60,000 of those half a million samples have tested positive so far and this is just Nigeria aggregate data The data that I’m presenting only applies to two states, or federal capital territory, so Kano and Abuja As they can see from our data, about 8,000 samples tested We only have seen a test positive rate of about 3%

in comparison to about 10% nationwide in Nigeria You can also see the timeline Testing started late April and May, already reached a peak in about September, and the numbers are now starting to go down for both total tests as well as positive This is the timeline for the percentage positive cases We can see that early on in the epidemic around June, May, we also saw about 10% test positive, right? But this has now tapered down to less than 3% at the moment We did some statistical tests for associations between symptoms as well as tests The symptoms and criteria on screen are taken directly from the Nigeria case investigation form For so for every sample collected, we collect certain symptoms as well as case definitions to give an indication as to the association, you can see most of these have shown positive association This is the breakdown by gender, total samples collected about 70% male, although when you look at the positive test about 74% are male This is the breakdown by age group You can see over the time course there has not been much change as far as the percentage of different age groups Most of our tests come from patients who are between 25 and 45 years of age So, that group accounts for about 50% of all of our samples Although when you look at the positive test you can see that the positive tests are slightly skewed towards the older age group And that’s 36 to 55 plus We are providing passenger screening services with PCR tests, like I mentioned before So far, we have tested about we have tested passengers from about 95 different countries The top five countries for incoming passengers to Nigeria are UK, India, Lebanon, China, and the U.S Interestingly, the top five countries that are recording positive cases from among the passengers are Russia, India, France, Kenya, and China The test positive rate that we see overall in the incoming passengers is less than 1%, compared to about 3% that we see in the general population With that I want to thank you for some of my colleagues for providing some of the data analysis, as well as the slides And thank you for listening – Thanks so much, Ravi. Very nice work Again, questions please into Q&A box I mean, they’ll be interesting I’m sure there are many questions here but for the sake of time, we move on to the last speaker, which is Frank Kagoro Frank is a Research Physician for the Epidemiology Department at Mahidol Oxford Tropical Medicine Research Unit, MORU, and he’s currently doing his PhD at the University of Cape Town His research focuses on tools for surveillance and mapping of infectious diseases and drug resistance Frank holds a doctor of medicine from Tanzania diploma in Healthcare Management and Tropical Countries from the Swiss Tropical and Public Health Institute and a master of science in International Health and Tropical Medicine from the University of Oxford With this please, go ahead Frank with your talk – Yes, so I’d like to share with you on user-centered dashboards for COVID-19 trends in Africa And I think if other members of the webinar now would appreciate is we have a lot of information that COVID-19 has offered us And there are a measurable number of ways we can measure the disease and tracking trends But the question comes to how much of this go to the users? And these are people who are on the frontline of the COVID response So, my work actually centers on how do we translate some of this knowledge and the measurements to the public Next, please And so the beginning of the pandemic in actually Africa we had a bit of a pause before the storm And everywhere COVID started in January but in Africa, somewhere around February That’s when we started experiencing cases in Sub-Saharan Africa countries towards the end of February And now Africa actually has a 4.2% of the global cases And if you compare that to the 15.7% of the population of what African population is to the world,

we might say somehow we are spared But then with the enormous amount of knowledge the challenge that presented in the beginning was there was lack of user-friendly and adaptive decision-making tools So we did a study which was more qualitative to speak to the frontline health-workers and what they would like to have so that they can be able to respond to COVID-19, next So, we decided to create up-to-date user-friendly COVID-19 visualizations And this was done to support this response, the responding teams, next So, one is what we call it, COVIDASH What COVIDASH does is, the aim is to develop and to share adaptive and interface that shows the burden of COVID-19 in Africa We can appreciate now there’s a lot of dashboards that have been put forward, with the Johns Hopkins dashboard leading the way, that they show in the other organizations However, in the beginning of the outbreak most of the dashboards focused on showing the burden at scientific level and a few of people who are on the frontline of response so, here we are talking about if we are talking about surveillance officers these are people who have basic education who go to root for COVID-19 cases for testing If it’s different teams, these are people who are trained to probably a simple high school level or bachelors, who don’t have an understanding of modeling So the aim was to simplify so that we can increase understanding and also to see the information that is being published So on the right side of the slides you will see that is the framework that we try to implement And below are the visualization that show of how this dashboard and I mean, the final product looks like, next I wish you can help me to play, – [Narrator] A video plays of Covidash. On a dashboard, 3 number values for Cases, Deaths, and Recoveries update in real time – Yes, so this is the COVIDASH What COVIDASH is, it’s a mobile friendly platform And it will show you where Africa world publications- I mean, burden, how the burden looks like, it will give you summaries, plots, table and maps It will give you traces of different places in Africa, top 10 cases and deaths It will give you Epi curves for different countries and 17 points where you can choose whether you want to see the whole of Africa, or just countries It will also show you a table of different countries and where they fall with different case fatality rates It also gives you a map where you can zoom in, show you cases / deaths and you can weigh them according by population to see where, I mean, when we are most affected by COVID And you’ll see South Africa, Egypt and Nigeria, which is so far currently show most of the cases So, this is part of what we thought could be user-friendly to healthcare workers and people who are in the frontline of COVID response So, all the models and work that we do, how best we can sort of tailor it to the end user, and they can be able to change and if they are different times how things work They can also see global statistic to just know a global summary of where we are at At the moment when we’re recording this, we’re still not passed to 1 million deaths And the most important part of COVIDASH it also gives you published information of COVID-19 in Africa So, this is a living systematically review that shows you what is information that is being collected- I mean, that’s been published for COVID-19 And you will see how studies envolved over time until we have more than a thousand of publications now on COVID-19 You can go on and look for every paper that has been published But another interesting part is where you will be able to evaluate any study and say which topic does it fall And when you click “submit” we receive this review, and it goes back into the table where someone if they don’t read their paper, like here they can select if they want to read surveillance system or treatment paper

So it helps them to use shorter time, instead of going to work in to look for more information anywhere, they can just get all of it just on their palm Next Let’s try it So, and they have a dashboard that we’re involved and this is part of the National COVID-19 Epi Model So, this is not the whole of Africa This is more of South Africa And this was to translate the outputs of the COVID-19 projection model to a user-friendly platform that is interactive It can be used up to the district level On the left bottom side, you’ll see that’s the framework So, people can, I mean, health-workers can interact with and the policymakers can interact with this dashboard to see the trends But also they can download the reports on the same framework on the right side you’ll see that it can also it also responds to frequently asked questions Next So, you can play just a little- I’ll speak a bit on the- this dashboard which is actually tailored for people who are working in the provinces and districts So, effective to have a disclaimer to tell them what projections are about And what are the data usage of this model It has projections which are cases, hospitalization and deaths So, all the models have already been run but the output that is given from the model, this is the way we thought we could present it for people to understand where we are going They can select the different cases whether severe or symptomatic and see how they’re projected over time They can open a table, they can also project it on a map, and this is for all the 52 districts of South Africa And also they can compare between different districts So, in this way like here, you can see we’re selecting different districts and one can compare also between different provinces And these are in cases, hospitalization, and also projected deaths And what it also gives you is, you can go back to time or you can go ahead to the future of the projections to see what we projected to see in the future with the trends of COVID-19 in South Africa It also gives you a resources page, which we can go into read more about the model, who are the collaborators, it gives you reports, also the frequently asked questions Where most of the questions which are support modeling, can also be found here, next So, a good thing with the strengths that have come from this work, next So, the strength is COVIDASH now is currently being used by clinicians for COVID-19 trends We get a lot of feedback of what we should be doing to improve It is low maintenance because it auto-updates itself It used R and R Shiny and data mining tools And for the South African dashboard, it’s been used by district teams to respond on COVID-19 and the software itself it’s been very useful, because all new data can be put in and projections can be shown So, it also now runs under low maintenance Also it gives PDF reports and the Excels So, the only limitations that come out of this work, is the challenge is all of these models require a lot of interpretation to be able to explain to people of what all this data that has been comes from all our analysis mean One challenge is we can be able to map COVID-19, but if we don’t release with, we don’t release this information for the public we fall into a lot of challenges So, we are limited on who to reach and how to reach them and also we had constraints with time of how to respond So, such platforms that are user-centered can be very useful And so there’s still a lot of opportunities, a lot of these tools are available online, and using open source coding, we can work globally and work with a lot of people

and so this is more like, with this kind of work we can foster collaborations that are global because some of the team are people who are based at somewhere in the world but we have been able to tailor things for African scientists and people who are responding So, with that I would like to say, thank you, next slide And apologies again for technical glitches So, one of the big challenges working also in our context is technology And I’m sorry, you experienced it for my presentation Thank you – [Narrator] Due to a connection issue the order of talks and Q&A had to be reversed Frank Kagoro could not take part in the Q&A – We’ve had very nice talks so far SO I think this will give us good material for discussion My co-moderator Cecile Viboud we will actually take over now, Cecile, are you online? Can you lead the Question Answering session? – Yes, of course So, we have quite a few questions and we’ll go through as many as possible A lot of those questions relate to the data that had been used in the different models So, we’ll go for that So, starting with Ali and the IHME model There’s a question about a serosurvey and that’s been over there are too few they are some one has been cited for Nairobi showing an 8.5% prevalence in June I’ve heard of some instances that show prevalence as high as 30% in rural areas And so those tend to be higher estimates than what the prediction of IHME model would portray So, Ali, I don’t know if you have any thoughts on how when you get more of those you could include them in your model – [Narrator] Ali Mokdad: – We include all the sero positive surveys So, the question said that is one that has 8%, I can’t recall exactly But some of them, how they do the testing we exclude them because of quality of testing or the type of testing But in Africa the question says it’s projected to be 40 We haven’t seen 40% in Africa, we don’t see it in South Africa, the highest we see right now And they asked us New Jersey 25, and Ecuador, there are surveys at 45, so I don’t think any country in Africa has a 40%, even 20% much lower than that, over – Yeah, so just to follow up on that a bit, it’s probably, I mean, there is supposedly a lot of heterogeneity in infection rate within a country, I mean, we see that in the U.S. right? And so presumably that’s also the case within each of those countries in Africa, how do you handle that in your model? – So, if we have a locality, one, we adjust if we take it and adjust for the population size and include it in the rest If it is, the key question for us to keep it or not to keep it, is what type of testing they did We use our labs here, we check with them if that’s a valid methodology so that our surveys in the United States in California that we didn’t include, for example – Great, okay, more questions for you So, you have interesting data on mask wearing in your model, how is that measured? – So, mask wearing and mobility is done by Google surveys And they have about 1 million every week and every country in the world except the like North Korea and couple of places But for Africa, which is a very good point to keep in mind, and that’s an overestimate of mask wearing because the more affluent are more likely to be on these smartphones and mobility is of course underestimated because the more affluent and more likely to respond and then more likely to stay at home But the source is Google and 1 million every week – Yeah, I mean, would you be thinking about potentially doing more representative survey just to try to titrate the Google data? – But IHME doesn’t do that but I actually take surveys that have been done in other location and calibrate So, we have switched for example, in the United States to another source, because all the other surveys independently, random surveys, were showing the numbers are different, so we switched to that We keep an eye on as a survey is one that available at the national level – we’d like to use Google but we adjust estimates if that’s available – Yeah, so more general question about data, data in Africa and how that collected So we understand that you’re not generating data yourself,

you’re collecting everything that’s available but rephrase that question a little bit Can you point at really key data gaps in some countries or types of data and then maybe alert the community that your estimates are perhaps not as viable in those places with less data? – It’s very good, very good question Yes, the quality of data varies from location and then certainly, I mean, to our surprise at IHME that in places, for example, in Africa some types of data is way better than everywhere else depending on what you’re collecting and what you’re doing So yes, we have a lot of gaps in Africa and data and vital legislation and the testing and mortality reporting we know there’ll be excess mortality we have no idea that there are a few locations that we aren’t very confident about in Africa So, there is a lot to be done, but in one of the questions, and I’m diverting a little bit but it’s still under the same, that I don’t believe that people said they don’t believe the data in Africa There is two points we need to keep in mind, the epediomiologist, the number itself, what’s the prevalence of people who are being tested who are positive for COVID-19 in the country? And there’s a trend for that And then if the quality of data is bad, but it’s done exactly the same way at every time, then the trend is more likely to be correct than that, so we need to keep that in mind All trends in Africa are showing a decline, so I have no reason not to believe there is a decline going on in Africa But yes, the number is an underestimate of course due to data quality and gaps in data, over – Great, so we’ll let you rest for just a second and move on to questions for Agnes So, another question about data quality, I suppose Do we really trust what you get on metrics on COVID cases and deaths? What is your opinion Agnes? – Yes, like the previous speaker has mentioned, data is always data The quality might be maybe compromisable at some point But in general, the epidemic shows that there’s a decline in some countries in the African continent and I think that on the Ugandan side the government has been very keen to increase testing we’ve seen the tests by thousand increasing from 0.01 to about 12, per 1,000 right now So, I guess with the prevalence we are seeing and the metrics, I think that it, we believe the data and with a decentralized testing and the ongoing quality control measures in the country to ensure that what is reported is correct I think that they- it’s the data is as is, and it might have big gaps but I think there’s a bigger part of it that is reliable – So, another question for you, Agnes, about, it’s more of a methodological question So, what tool did you use for your sentiment analysis and the values of the viable trust, mistrust, etcetera, that we saw on your graph, what’s that based on? – Okay, so the analysis is done using Python and R and the different algorithms and the values would show if it’s a negative sentiment It’s any value that’s negative or less than zero is a negative sentiment and the positives or those above one and above zero are positive sentiment But we did that using Python – Great, one more methodological question How did you identify Twitter influences is that like a network model? – Yes, it’s sort of the network analysis and we largely relied on the tweets and their retweets And if you conduct that analysis and you find that there are people who have- there’s some kind of centrality around them, you find that that person has- is already tagged in a tweet or retweet, then we used how many nodes were- how big, how close that person wants, or how deep their nodes was in the network analysis, and then we call them influencers But this is still ongoing work – Okay, great And just to reassure you, can you confirm that when you get data from Twitter, it is anonymized? – Yes – Okay, so I am told that we can resume the talk session with our last speaker and then if there is any time we’ll take a few more questions at the very end Thanks, over – So we still have a few questions, but you can ask those in the breakout sessions and we’ll give you some guidelines for that just but just to wrap up It was a really good set of diverse talk describing both burden estimates, but also projections for COVID-19

And we touched on how difficult the data are which is not a problem that’s specific to Africa We, everyone around the world is struggling with COVID-19 information but that’s also an opportunity to use novel data streams as we’ve seen, right, including Google and Twitter, and also interesting tool to help with monitoring of suspected cases and returning travelers And also the importance as we’ve seen in the last talk of just visualizing this data and making them publicly available to the public and understandable as well So, great session, there’s more opportunities for discussion in smaller groups where we can see all of the participants with the panelists So, first session on direct impact of COVID-19, with Ali and Elaine and myself The second session on indirect impact, we’ve heard about HIV, but also as of question about impact on childhood infections and vaccination, so, that’s with Agnes and Amit And then another session on methods with Stefan, Ravi and Frank And I’ll hand it over to Puleng who’ll give us more guidelines on how exactly to reach those sessions, over – Fantastic, thank you so much Cecile, and thank you, Stefan and fantastic moderators Thank you, special thanks to you, Frank, for just bouncing back and being able to give those, that talk in the midst of the technology challenges I’m based in South Africa so I fully relate to the technology challenges So, thank you so much for still being able to present to us and to every panelist who has shared their presentation, their wisdom and their very interesting and timely discussions, we really appreciate you Now, I would like at this point to invite panelists and moderators to go ahead into the breakaway rooms, get ready and set it up for the interactive sessions with the participants who would like to engage with you further on the three topics that we will highlight And for all the participants, I invite you to just stay in one more moment, to continue a bit of conversation with you, about how to join that interactive post-webinar program But first, in the chat, we’re about to put in that link and it’s also on the screen and that’s an exit poll, – [Narrator] Go.hub.KI/SoDSFightingCovid – and it’s also on the screen and that’s an exit poll, we would like you to please click through it and give us feedback very quickly, 20 seconds of your time We’re actually going to give you some time to do it right here Please click on the link and give us some feedback about today’s session and what has worked really well for you and what hasn’t worked We really appreciate your feedback We appreciate your attention today, thank you for joining us and we’re going to give you a couple of seconds to fill the survey So, click on the link and you should be able to get a couple of questions that we’d like to invite you to engage with just for 20 seconds Great, so we do have an another upcoming session in the next week So please do check out the screen We have placed the title of the upcoming session in this series So, the program is going to be on Health Metrics That’s also going to happen on the 21st of October on Wednesday, – [Narrator] Wednesday October 21st. Innovations in Health Metrics Sciences: Measuring, Mapping and Monitoring Morbidity and Mortality at the Regional, National and Local Levels in Africa – please do come and visit us and spend time with us Same time, same place, and go to KIStorm, and you will get the link to the webinar and the interactive session Thank you so much Now, are you eager on how to get into the interactive session which is a really close interaction, with the panelists today? All of those questions that we didn’t manage to answer on the webinar, you can ask them directly to the panelists in an intimate setup where is just the panelists and yourselves and you can engage and you can direct the conversations however way you would like to So, in order to get there, the same link that you used to get here today,

is the same link that’s going to lead you to the interactive session So, when you do leave this room and returned to your browser it would have refreshed with several virtual Post-It notes on your screen that will appear on this page as you scroll down And one for each of the four breakaways, the three breakaway sessions, that we’ll have The fourth breakaway session will be just for the technical assistance So, each of them- breakaway rooms has its own Zoom Room, you will click and on “Join the Zoom”, to enter the interactive session, and that will last about 30 minutes You might find yourself really interested in one topic and you can stay there for the whole time, otherwise, you’re welcome to visit the rest of the Zoom Rooms by going to the same link Remember that if you need help at all or get lost or get bumped out of a session or lose your way or just find just find the “Team Help” on the KISTorm page button And we’ll give you the help that you need to get to the breakaway session With that, I just want to say thank you to the moderators who have done a fantastic job today and our panelists, and most of all, to all of you for being with us, until this far. We hope to see you again, shortly in a few moments in those interactive sessions and next week at the State of Data Science Session on Health Metrics Thank you so much We’ll see you in the breakaway rooms Thank you – [Narrator] Thank you for participating in this DSI Africa State of Data Science session! Visit KI Storm for upcoming sessions (upbeat music)