Accelerating AI for global health through crowdsourcing

Curated on Posted on by Stefaan Verhulst

Poster by Geoffrey Henry Siwo: The promise of artificial intelligence (AI) in medicine is advancing rapidly driven by exponential growth in computing speed, data and new modeling techniques such as deep learning. Unfortunately, advancements in AI stand to disproportionately benefit diseases that predominantly affect the developed world because the key ingredients for AI – computational resources, big data and AI expertise – are less accessible in the developing world. Our research on automated mining of biomedical literature indicates that adoption of machine learning algorithms in global health, for example to understand malaria, lags several years behind diseases like cancer.


 To shift these inequities, we have been exploring the use of crowdsourced data science challenges as a means to rapidly advance computational models in global health. Data science challenges involve seeking computational solutions for specific, well-defined questions from anyone in the world. Here we describe key lessons from our work in this area and the potential value of data science challenges in accelerating AI for global health.


In one of our first initiatives in this area – the Malaria DREAM Challenge – we invited data scientists from across the world to develop computational models that predict the in vitro and in vivo drug sensitivity of malaria parasites to artemisinin using gene expression datasets. More than 360 individuals drawn from academia, government and startups across 31 countries participated in the challenge. Approximately 100 computational solutions to the problem were generated within a period of 3 months. In addition to this sheer volume of participation, a diverse range of modeling approaches including artificial neural networks and automated machine learning were employed….(More)”.

Winter is coming. Can cities use innovation to save ‘streateries’?

Curated on Posted on by Stefaan Verhulst

Bloomberg Cities: “Outdoor dining has been a summer savior in these COVID times, keeping restaurants and the people they employ afloat while bringing sidewalks and streets once hushed by stay-at-home orders back to life.

But with Labor Day now behind us, many city leaders and residents alike are asking, “What’s next?” “What becomes of the vibrant ‘streateries’ once winter comes rolling in?”

Perhaps it’s no surprise that Chicago, notorious for its frigid winters and whipping lakefront winds, is at the forefront of the hunt for an answer. The city recently launched the City of Chicago Winter Dining Challenge to get everyone from designers to dishwashers thinking up new ideas for how to do outdoor eating in the cold in a way that is both appealing and safe for customers and restaurant workers.

More intriguing is just how much interest the competition has generated, including nearly 650 entries from all over the world. There are dozens of takes on warming large patios and small dining pods, including approaches likened to greenhouses, igloos, and yurts; ideas for repurposing parking garages and city buses; furniture-based concepts with heated tables, seats and umbrellas, and even a Swiss-style fondue chalet.

The goal, said Samir Mayekar, Chicago’s Deputy Mayor for Economic and Neighborhood Development, is to surface ideas city leaders would never have thought of. Three winners will get $5,000 each and see their ideas piloted in neighborhoods across the city in October….(More)”.

The Principle of Self-Selection in Crowdsourcing Contests – Theory and Evidence

Curated on Posted on by Stefaan Verhulst

Paper by Nikolaus Franke, Kathrin Reinsberger and Philipp Topic: “Self-selection has been portrayed to be one of the core reasons for the stunning success of crowdsourcing. It is widely believed that among the mass of potential problem solvers particularly those individuals decide to participate who have the best problem-solving capabilities with regard to the problem at question. Extant research assumes that this self-selection effect is beneficial based on the premise that self-selecting individuals know more about their capabilities and knowledge than the publisher of the task – which frees the organization from costly and error-prone active search.

However, the effectiveness of this core principle has hardly been analyzed, probably because it is extremely difficult to investigate characteristics of those individuals who self-select out. In a unique research design in which we overcome these difficulties by combining behavioral data from a real crowdsourcing contest with data from a survey and archival data, we find that self-selection is actually working in the right direction. Those with particularly strong problem-solving capabilities tend to self-select into the contest and those with low capabilities tend to self-select out. However, this self-selection effect is much weaker than assumed and thus much potential is being lost. This suggests that much more attention needs to be paid to the early stages of crowdsourcing contests and particularly to those the hitherto almost completely overlooked individuals who could provide great solutions but self-select out.”…(More)”.

The detection and location estimation of disasters using Twitter and the identification of Non-Governmental Organisations using crowdsourcing

Curated on Posted on by Stefaan Verhulst

Paper by Christopher Loynes, Jamal Ouenniche & Johannes De Smedt: “This paper provides the humanitarian community with an automated tool that can detect a disaster using tweets posted on Twitter, alongside a portal to identify local and regional Non-Governmental Organisations (NGOs) that are best-positioned to provide support to people adversely affected by a disaster. The proposed disaster detection tool uses a linear Support Vector Classifier (SVC) to detect man-made and natural disasters, and a density-based spatial clustering of applications with noise (DBSCAN) algorithm to accurately estimate a disaster’s geographic location. This paper provides two original contributions. The first is combining the automated disaster detection tool with the prototype portal for NGO identification. This unique combination could help reduce the time taken to raise awareness of the disaster detected, improve the coordination of aid, increase the amount of aid delivered as a percentage of initial donations and improve aid effectiveness. The second contribution is a general framework that categorises the different approaches that can be adopted for disaster detection. Furthermore, this paper uses responses obtained from an on-the-ground survey with NGOs in the disaster-hit region of Uttar Pradesh, India, to provide actionable insights into how the portal can be developed further…(More)”.

Disinformation Tracker

Curated on Posted on by Stefaan Verhulst

Press Release: “Today, Global Partners Digital (GPD), ARTICLE 19, the Collaboration on International ICT Policy for East and Southern Africa (CIPESA), PROTEGE QV and  the Centre for Human Rights of the University of Pretoria jointly launched an interactive map to track and analyse disinformation laws, policies and patterns of enforcement across Sub-Saharan Africa.

The map offers a birds-eye view of trends in state responses to disinformation across the region, as well as in-depth analysis of the state of play in individual countries, using a bespoke framework to assess whether laws, policies and other state responses are human rights-respecting. 

Developed against a backdrop of rapidly accelerating state action on COVID-19 related disinformation, the map is an open, iterative product. At the time of launch, it covers 31 countries (see below for the full list), with an aim to expand this in the coming months. All data, analysis and insight on the map has been generated by groups and actors based in Africa….(More)”.

How Crowdsourcing Aided a Push to Preserve the Histories of Nazi Victims

Curated on Posted on by Stefaan Verhulst

Andrew Curry at the New York Times: “With people around the globe sheltering at home amid the pandemic, an archive of records documenting Nazi atrocities asked for help indexing them. Thousands joined the effort….

As the virus prompted lockdowns across Europe, the director of the Arolsen Archives — the world’s largest devoted to the victims of Nazi persecution — joined millions of others working remotely from home and spending lots more time in front of her computer.

“We thought, ‘Here’s an opportunity,’” said the director, Floriane Azoulay.

Two months later, the archive’s “Every Name Counts” project has attracted thousands of online volunteers to work as amateur archivists, indexing names from the archive’s enormous collection of papers. To date, they have added over 120,000 names, birth dates and prisoner numbers in the database.

“There’s been much more interest than we expected,” Ms. Azoulay said. “The fact that people were locked at home and so many cultural offerings have moved online has played a big role.”

It’s a big job: The Arolsen Archives are the largest collection of their kind in the world, with more than 30 million original documents. They contain information on the wartime experiences of as many as 40 million people, including Jews executed in extermination camps and forced laborers conscripted from across Nazi-occupied Europe.

The documents, which take up 16 miles of shelving, include things like train manifests, delousing records, work detail assignments and execution records…(More)”.

Saving Our Oceans: Scaling the Impact of Robust Action Through Crowdsourcing

Curated on Posted on by Stefaan Verhulst

Paper by Amanda J. Porter, Philipp Tuertscher, and Marleen Huysman: “One approach for tackling grand challenges that is gaining traction in recent management literature is robust action: by allowing diverse stakeholders to engage with novel ideas, initiatives can cultivate successful ideas that yield greater impact. However, a potential pitfall of robust action is the length of time it takes to generate momentum. Crowdsourcing, we argue, is a valuable tool that can scale the generation of impact from robust action.

We studied an award‐winning environmental sustainability crowdsourcing initiative and found that robust action principles were indeed successful in attracting a diverse stakeholder network to generate novel ideas and develop these into sustainable solutions. Yet we also observed that the momentum and novelty generated was at risk of getting lost as the actors and their roles changed frequently throughout the process. We show the vital importance of robust action principles for connecting ideas and actors across crowdsourcing phases. These observations allow us to make a contribution to extant theory by explaining the micro‐dynamics of scaling robust action’s impact over time…(More)”.

UK parliamentary select committees: crowdsourcing for evidence-based policy or grandstanding?

Curated on Posted on by Stefaan Verhulst

Paper by the The LSE GV314 Group: “In the United Kingdom, the influence of parliamentary select committees on policy depends substantially on the ‘seriousness’ with which they approach the task of gathering and evaluating a wide range of evidence and producing reports and recommendations based on it. However, select committees are often charged with being concerned with ‘political theatre’ and ‘grandstanding’ rather than producing evidence-based policy recommendations. This study, based on a survey of 919 ‘discretionary’ witnesses, including those submitting written and oral evidence, examines the case for arguing that there is political bias and grandstanding in the way select committees go about selecting witnesses, interrogating them and using their evidence to put reports together. While the research finds some evidence of such ‘grandstanding’ it does not appear to be strong enough to suggest that the role of select committees is compromised as a crowdsourcer of evidence….(More)”.

Adaptive social networks promote the wisdom of crowds

Curated on Posted on by Stefaan Verhulst

Paper by Abdullah Almaatouq et al: “Social networks continuously change as new ties are created and existing ones fade. It is widely acknowledged that our social embedding has a substantial impact on what information we receive and how we form beliefs and make decisions. However, most empirical studies on the role of social networks in collective intelligence have overlooked the dynamic nature of social networks and its role in fostering adaptive collective intelligence. Therefore, little is known about how groups of individuals dynamically modify their local connections and, accordingly, the topology of the network of interactions to respond to changing environmental conditions. In this paper, we address this question through a series of behavioral experiments and supporting simulations. Our results reveal that, in the presence of plasticity and feedback, social networks can adapt to biased and changing information environments and produce collective estimates that are more accurate than their best-performing member. To explain these results, we explore two mechanisms: 1) a global-adaptation mechanism where the structural connectivity of the network itself changes such that it amplifies the estimates of high-performing members within the group (i.e., the network “edges” encode the computation); and 2) a local-adaptation mechanism where accurate individuals are more resistant to social influence (i.e., adjustments to the attributes of the “node” in the network); therefore, their initial belief is disproportionately weighted in the collective estimate. Our findings substantiate the role of social-network plasticity and feedback as key adaptive mechanisms for refining individual and collective judgments….(More)”.

Using crowdsourcing for a safer society: When the crowd rules

Curated on Posted on by Stefaan Verhulst

Paper by Enrique Estellés-Arolas: “Neighbours sharing information about robberies in their district through social networking platforms, citizens and volunteers posting about the irregularities of political elections on the Internet, and internauts trying to identify a suspect of a crime: in all these situations, people who share different degrees of relationship collaborate through the Internet and other technologies to try to help with or solve an offence. T

he crowd, which is sometimes seen as a threat, in these cases becomes an invaluable resource that can complement law enforcement through collective intelligence. Owing to the increasing growth of such initiatives, this article conducts a systematic review of the literature to identify the elements that characterize them and to find the conditions that make them work successfully….(More)”.