The new informatics of pandemic response: humanitarian technology, efficiency, and the subtle retreat of national agency

Diagnostics

This first component groups diagnostic activities related to collecting, systematizing, and processing information about a disease outbreak, and mapping its spread and associated needs, which we term diagnostics. This constitutes diagnosis at a societal level, and should not be confused with the medical diagnosis of individuals.

The analysis of information has been central to understanding, anticipating and responding to infectious diseases at least since John Snow became the “father of modern epidemiology” by drawing dots around a map of London water pumps during the Cholera epidemic in 1854 (Hempel 2006). In the humanitarian context, collecting information is central to the implementation of an efficient response, including situational information, needs assessment, and operational information (King 2005; Van de Walle et al. 2008). This corresponds with humanitarian practice which has traditionally assumed the “information imperative” to be central to the humanitarian imperative—that is, the need to collect as much relevant information as possible, to enhance evidence for decision-making basis and improve efforts to assist people suffering in crisis contexts (Bui et al. 2000; Darcy and Hofmann 2003; Miller et al. 2005; Saab et al. 2008).

Scholars have explored a number of ways in which technology improves humanitarian diagnostics, including the use of mobile phone network data for human mobility mapping and contact tracing (Tatem et al. 2009; Aslam et al. 2014; Wesolowski et al. 2014; Gittelman et al. 2015; Bharti et al. 2015; Bengtsson et al. 2015), and big data or social media scraping for contagion modeling (Brownstein et al. 2009; Chunara et al. 2012). Such techniques allow humanitarian actors to extract information, diagnosing needs and epidemiological trends without direct contact with affected populations.

Other applications of digital media for diagnostic activities introduce completely different actors. Novel initiatives in response to the Ebola pandemic in 2014–2016 include hackathons organized in western capitals to map resources for West African response, and western non-governmental organizations (NGOs) that provide communication-based outbreak models to multinational humanitarian organizations (Sangokoya 2014; Moore 2015; Dittus et al. 2016) or build geographic information system (GIS) maps for national authorities (Timo Lüge 2014). Early work is even underway to develop artificial intelligence responses to combat the spread of infectious diseases, by using multiple sources of publically available data to algorithmically predict the appearance and spread of disease (Barron 2014).

Such innovative approaches are compelling and have attracted an understandable amount of interest and optimism, not in the least due to perceived gains in efficiency. Novel collaborations and web-based health surveillance systems have regularly been quicker than the World Health Organization (WHO) to publish reports on epidemic outbreaks (Anema et al. 2014: 1036-1037), and Jennifer Gardy, a senior scientist at the British Columbia Centre for Disease Control, has argued that epidemics are always essentially a race between the spread of info and spread of virus, and that technology has given scientists a critical edge (Edmunds 2013).

Whether or not such approaches are inherently more efficient, they are noteworthy for their inclusion of novel actors. This is perhaps most apparent in web-based initiatives such as wiki systems, which are used to facilitate collaborative pandemic modeling among geographically disparate scientists (Kno.e.sis 2014), or the Humanitarian Data Exchange, which through the leadership of a traditional humanitarian agency, was used to coordinate data sharing between a diverse group of actors, including local NGOs and remote volunteer mapping communities (Verhulst 2014; Fast and Waugaman 2016: 90). Though not solely dependent on digital media, a similar dynamic is visible in the proliferation of global health networks, such as GOARN (Global Outbreak Alert and Response Network) and ProMED (Program for Monitoring Emerging Diseases), which combine the efforts of national and international civil society and regulatory bodies, to surveil health risks internationally through a mix of traditional and digital media, and which to some degree eclipse the traditional roles of international and national health regulatory bodies (Ramalingam 2015: 9–13).

Risk communication

The interactive character referenced in that definition is the subject of some debate. In their Handbook on Global Health Communication, editors Waisbord and Obregon argue that the field of global health communication is characterized by a theoretical split, which places a preference for behavior change and unidirectional approaches against critical theories of participatory engagement (Waisbord and Obregon 2012: 7–33). Civil society and practitioner rhetoric tends to link participatory communicative models with technological advances, framed in normative terms that are highly critical of traditional, less participatory approaches in the humanitarian and development sectors (Chao 2014). This normative perspective is also assumed by several scholars (Kaiser 2000; Abraham 2011; Maxwell et al. 2011; Gillman 2014; Özdamar and Ertem 2015; Madianou et al. 2016), and participatory approaches to risk communication have been argued to improve outcomes and efficiency. Odugleh-Kolev (2014) argues that a structural and interactive understanding of risk communication is particularly important in pandemic response, where coordinated, functional, and systemic communication is implied in activities such as assessing transmission risk and mapping contagion patterns in rural communities (243). Participatory approaches to risk communication tend to incorporate diagnostic activities, since digital media technologies facilitate simultaneous broadcast and collection of humanitarian information, at marginal cost, as will be discussed below.

For clarity, this analysis focuses on risk communication implemented nationally or sub-nationally, targeting communities potentially affected by pandemics, and disregard the international role of for instance social media in risk communication. In doing so, we can identify at least three ways in which digital media are used intentionally to bolster risk communication strategies. Again, responses to the Ebola outbreak of 2014 are illustrative:

Firstly, digital media was expected to expand the reach and interactivity of risk communications. The use of new media technologies, particularly mobile phones, is often expected to dramatically increase the reach of risk communication to rural and remote communities, as well as front line health providers. Mobile penetration rates in developing countries were estimated to have surpassed 90% at the time of the Ebola outbreak (Wesolowski et al. 2014), and a review of Ebola response in Nigeria cited mobile-delivered training to health workers as “vital” for promptly declaring the country “free of Ebola” (West 2015: 10). Digital media also allow for scaled co-creation of health communication content, such as programs to produce videos together with youth infected with Ebola in Sierra Leone, as a means of addressing stigmatization and developing trust networks among affected populations (Zuckerman 2014).

Secondly, digital media allow for feedback and interaction to be integrated into risk communications. A project called U-Report, led by the United Nations Children’s Fund (UNICEF), combines mobile phone-based surveys with opportunities for user feedback and was used to channel infection reports and concerns regarding public health activities in Liberia during the Ebola outbreak (Muah et al. 2014). Similarly, the international conglomerate IBM combined outgoing radio communications with interactive short messaging system (SMS) functionality to implement interactive risk communications surrounding Ebola in Sierra Leone (Bell, 2014). Some studies suggest that these affordances were particularly effective when targeting health communication to specific sub-groups (Ems and Gonzales 2015).

Finally, digital media are sometimes expected to obviate institutional and resource limitations on the development of media platforms and content production for public health authorities. That digital media is faster and cheaper to produce and disseminate than paper is widely recognized. In addition, participatory content creation such as the Sierra Leone video program described above can effectively outsource some degree of communications work, as do content creation efforts which move beyond specific communities and aim to engage “the crowd” in generating content. Such approaches have been pursued in generating both platforms and content for risk communication, either through international volunteer communities or institutionalized global health networks, or initiated by third parties, as in a series of hackathons organized in New York to develop mobile apps for spreading information about Ebola in affected countries (Sangokoya 2014).

Coordination

The third component we identify is coordination, which is a persistent challenge and polemic in the humanitarian sector (Stephenson 2005; Bisri 2016). The introduction of new actors and new technologies promise to mitigate this challenge, even as they contribute to it in novel ways. Ramalingam (2015) notes that the contemporary governance system for international diseases incorporates at least five different types of actors (Intergovernmental organizations, National governmental organizations, Non-governmental organizations, Private foundations, and Public/private partnerships and consortia), whose activities are marked by competition and lack of collaboration (ibid 5–6). He argues that this institutional disarray, when coupled with other socio-economic developments and trends of globalization, poses a threefold governance challenge to global health: the challenge of capitalizing on the diversity of actors, the challenge of bringing traditional actors up to speed with innovative organizational and technical approaches, and balancing the incentives and responsibilities of individual countries in treating diseases that do not respect borders (ibid 6).

Simultaneously, timely, accurate, and appropriate information is widely regarded as a cornerstone for effective humanitarian coordination (Bui et al. 2000; Miller et al. 2005; Saab et al. 2008) and digital media are often expected to dramatically enhance the coordination potential of information (Moss and Townsend 2006). The most remarkable applications of digital media to coordination informatics for pandemic response are likely the creation and activation of global health networks, mandated to surveil infectious diseases internationally and alert regarding their outbreak. These networks have at least partially filled a coordination gap for identifying and initiating responses to global pandemics (see Burkle et al. 2012), but coordination in the implementation of specific response remains largely lacking.

Volunteer and Technical Communities (V&TCs) represent another prominent example of the coordination challenges posed by novel actors in humanitarian response. Volunteer communities are composed of thousands of individuals around the globe who are moved to contribute time and energy to humanitarian response efforts virtually and remotely, often by collecting or processing humanitarian information such as incident reports or spatial data over social media. Though different communities vary significantly in their membership, degree of organization and formal relationships with humanitarian coordinating bodies, their integration into a field marked by seasoned professional field staff and conservative information management is consistently marked by cultural and institutional tensions (Harvard Humanitarian Initiative 2011). Officially and institutionally this is visible in the challenges that surround the development of ethical codes of conducts for digital humanitarians (Meier 2015: 45-60; Resor 2016) or the development of an activation protocol through which OCHA is to include the Standby Task Force (SVT, one of the most prominent V&TCs) in humanitarian response operations (Burns 2014; Gorp 2014).

In Ebola response, V&TCs were particularly prominent in establishing novel information exchange platforms, such as a skype channel for coordinating data collection and identifying data gaps (Fast and Waugaman 2016: 84–88), though there is some evidence that lack of integration into formal coordination mechanisms actually “contributed to gaps in awareness of existing tools and duplication of effort” (ibid 86). This dynamic recalls of other instances where the introduction of digital tools complicates rather than facilitates humanitarian efforts (Bui et al. 2000; Miller et al. 2005; Saab et al. 2008), and is consonant with the more general assessment that the uncoordinated introduction of novel actors into Ebola response exemplifies “wider dysfunction in the provision of global health security” (Harman and Wenham 2018: 10).

Tapia et al.’s (2012) two case studies on humanitarian coordination offer a possible explanation for the failure of information systems to lead to enhanced coordination despite explicit efforts. On the basis of a careful literature review and analysis of two humanitarian coordinating bodies, they examine “the instrumental use of IT as a mechanism by which NGOs collaborate” (ibid 253) and identify an important distinction between IT-driven coordination efforts that are conceptualized as technical or informational challenges, and those that are conceptualized as organizational or process challenges, which they describe as more formidable. Tapia et al. conclude that coordination efforts with modest goals and modest demands on organizational processes are likely to be more successful and increase opportunities for successively more progressive coordination efforts (ibid 253). According to this logic, it would be reasonable to expect that digital media coordination would be most successful when it did not demand changes in organizational practice (i.e., coordinating actors who are already inducted in the use of specific media and coordination efforts that do not require changes to data formats or data collection procedures).

Influx of novel actors

New informatics introduce a host of actors and intermediaries not traditionally included in pandemic response and coordination. The ways in which this influx interacts with traditional structures for pandemic response can be considered according to a sandwich model, in which interaction is introduced from above and from below. The bottom slice of the sandwich in this metaphor is interaction with affected populations, where novel informatics present at least two types of challenges.

Firstly, social media and big data introduce promising new sources of information on which to base decision-making in pandemic response, but for whose meaningful use humanitarian organizations tend to lack the institutional and technical capacity, and national authorities even more so (Harvard Humanitarian Initiative 2011; Odugleh-Kolev 2014; Smith 2015; Read et al. 2016). Simultaneously, the participatory ethos of new technology encourages humanitarian organizations and government authorities alike to deliberately engage affected communities in the design, implementation and evaluation of humanitarian response (Kaiser 2000; Maxwell et al. 2011; Gillman 2014; Özdamar and Ertem 2015), yet poses a number of non-trivial hurdles to meaningful engagement. Here too, technical capacities tend to be weakest with national governments, necessitating partnership with international organizations to invest in participatory interventions. Infrastructural requirements and issues of access and representativity can also frustrate intentions to utilize participatory technologies. In a humanitarian context, political realities can often be the most meaningful obstacle, even when all issues of capacity, infrastructure, and access are surmountable. For all these reasons, national authorities are rarely in a position to unilaterally dictate the ways in which new technologies are leveraged to interact with affected populations.

Informatic challenges at the top end of the sandwich arise from the (sometimes unsolicited) engagement from novel international actors. Of particular note are V&TCs and digitally native civil society organizations that are small, nimble, and eager to disrupt established practice. These actors present fundamental challenges to humanitarian coordination by their very engagement. Because they do not fit neatly into traditional humanitarian coordination mechanisms, yet tend to demand attention and heighten expectations, novel international actors at the top level of pandemic informatics are a powerful force for asserting knowledge politics in informatics of response. The information flows described by Fast and Waugaman illustrate not only information exchanges in this sense, but fundamental assertions and negotiations about what kind of information is relevant and where and by whom those decisions are made. The moments at which this happens are not regulated by traditional policies of humanitarian coordination or cluster mechanisms, but occur in the practical application of technologies to knowledge and information, what some scholars have termed moments of closure in humanitarian knowledge politics (Burns 2014).

In some instances, international informatics are efficiently structured to serve national authorities. The introduction of data and information clearing houses presents opportunities for national authorities to assert control of national agendas, for example, such as when the Humanitarian Data Exchange enabled Guinean ministries to track training of infection prevention and control training efforts in the country (Fast and Waugaman 2016: 90). This dynamic appears exceptional in a chaotic response environment marked by a “myriad of actors with no clear role or leadership” (Harman and Wenham 2018: 10), however. A pandemic response context where “many of the information collection systems that organizations set up during the response were not linked to national systems or national capacity” (56) necessarily reinforces the capacity and agency of international actors, and often novel actors, at the expense of national authorities’ influence over response processes.

It is also worth noting informatics practices that transcend this two-level model. Most clearly in opposition to the agency of national authorities is the introduction of “hidden actors”, such as application developers, producers of hardware or network managers with de facto influence over humanitarian data during its collection and processing (see Gillman 2014: 7). Equally notable are instances in which international actors bypass engagement with the traditional response environment altogether. In some instances, this appears to occur exclusively at the international register, with little or no contact with any in-country actors. Examples include international networks of scientists collaborating to improve diagnostic tools (Eclipse 2017), or hackathons organized in foreign capitals to develop data models or applications for implementation in pandemic response (Sangokoya 2014; Gordon 2016; Lodato and Di Salvo 2016). Such initiatives raise serious questions about opportunity cost and efficient use of resources.

To summarize, digital communication technologies and information flows introduce a host of novel actors to pandemic response. Traditional humanitarian actors may experience this at the top level through the novel engagement of international actors, or at the bottom level through pressure to engage with affected populations and their data. In each instance, the influx of novel actors carries a risk for decision-making authority and agency to be moved into novel fora, further outside the influence of national authorities.

This may not occur in an absolute sense; indeed, the role of national governments in driving Ebola response strategies has been significant (DuBois et al. 2015: 21). Yet the potential for exclusion merits careful consideration, particularly given the already pronounced tendency for agency of national authorities to be limited by the militarization of humanitarian response (Sandvik 2015) and the increasing prominence of international NGOs in public health service provision (Prince 2014). Such dynamics are particularly vulnerable in the contexts of pandemic and crisis response, and reviews of the Ebola response have also noted how decision-makers’ identities shape response strategies (DuBois et al. 2015: 27).

The application of digital technologies has undoubtedly opened up a host of opportunities for decision-making by novel actors. This may occur in situations where national authorities enjoy a limited role, such as the network communications of global watchdog networks in which NGOs, scientific communities, international health networks and government agencies contribute to multiple digital communications streams and daily webinars to coordinate disease surveillance and response (Ramalingam 2015: 11–13). It may also occur in digital for where there is no direct participation by national authorities, or in situations where national authorities lack the basic technical capacities to engage in a natural coordinating role, such as when “a host of academics, private philanthropists and technology companies” lobbied telecom companies for access to call detail records in order to develop their own response strategies (Sandvik et al. 2017: 16).

Though there are some cases in which novel actors and information practices support a stronger role for national authorities in pandemic response and a greater capacity to exercise agency and decision-making in the design and implementation of that response (Ramalingam 2015: 10-14; Fast and Waugaman 2016: 63-66), but this appears rare and the conditions under which it occurs are unclear. The overwhelming picture is one in which technologically driven informatics exacerbate coordination challenges (Kim 2014), driving the enactment of knowledge politics outside formal structures of humanitarian clusters and beyond the influence of national authorities.

Convergence

Though the boundaries between diagnostic activities, risk communication, and response coordination were perhaps never entirely clear, these areas have traditionally been institutionally and procedurally distinct. Digital communications’ affordances and functionalities make it increasingly possible to combine activities from these areas in single processes.

Notably, the diverse communications between national authorities, WHO staff, NGOs, and scientific institutions are consolidated in a coherent institutional framework and through weekly webinars. This process interoperates diagnostic and coordination, feeding them directly into the establishment of response protocols, including protocols for risk communication, which are in many instances executed by the same actors that provided diagnostic information, by virtue of their capacity for rapid, two-way communication with populations.

Though it is unclear the degree to which this operational convergence of such tasks diminishes the agency of national authorities in a general sense, it is reasonable to expect that the inclusion of multiple actors decreases governments’ scope for top down control. This may often be for the better in terms of effective response, as is likely the case with SARS in 2013, when GOARN’s access to digital communications platforms facilitated the supply on non-governmental diagnostic information, which likely contributed to acknowledgement of the outbreak by the Chinese government.

Assisted Contact Tracing provides a second example. Contact tracing is the epidemiological practice of identifying the individuals who have come into contact with infected individuals, in order to map the spread of a disease (Huerta and Tsimring 2002). In 2014, a private company named Odisi developed a platform for “Assisted Contact Tracing” (ACT), which digitized this process through the use of Integrated Voice Recognition Software. Individuals in Ebola-affected communities were able to report their contacts using mobile phones and then received follow-up messages regarding care and updates on the Ebola response. This digitized approach increased the efficiency of data collection by eliminating the need for human interviewers, and also allowed the integration of other types of data (paper and mobile data), which increased the platform’s diagnostic capacity dramatically. Though the platform was designed and implemented as a diagnostic tool, the affordances offered by digital media quickly presented other opportunities. The automated registry of exposed individuals was quickly adapted for risk communication purposes via SMS follow-up messages, and the digitization of rapid analysis of data promptly positioned the ACT platform to play a coordinating role among parallel diagnostic initiatives. Here, we see a clear digital convergence of diagnostic and risk communication activities on a mobile phone platform, at the discretion of a private company.

A third example is offered by U-Report, a free SMS-based polling tool launched in Uganda in 2011 through a civil society partnership in order to monitor the quality of human rights and governance in the country, with a focus on polls related to human rights. In early 2012, platform users noted early signs of the outbreak that would later come to be known as “nodding sickness” and would claim over 3000 lives in the following months. U-report did not have a health mandate and did not solicit these early epidemic reports, but received them because the communication platform was already in place and integrated into the communication habits of users. In this sense, the presence of a digital media infrastructure very much conditioned the implementation of a diagnostic tool, which promptly provided a site for national coordination, as U-Report collaborated with the Ministry of Health and the WHO to develop and implement a 4-stage communications and mobilization plan. Here again we see the digital and operational convergence of diagnostic activities and risk communication activities by national authorities with the support of international organizations. Notably, while national authorities are directly engaged in both sending and receiving content, they do not have direct control of the operational and financial processes that support the initiative.

Fourth and lastly, the Humanitarian Data Exchange (HDX) was established by the UN Office of Coordination for Humanitarian Affairs in 2014, in an effort to improve and coordinate access to humanitarian data. The Ebola Crises Page collected 62 data sets from UN, governmental, civil society, and private sector data sources and invited users to contribute their own data to the site. The page also featured maps and visualizations developed in collaboration with private charitable foundations and private businesses. As such, the site represents a near seamless integration of all the components of pandemic informatics. It directly served the needs of independent diagnostic efforts by providing access to quality humanitarian data; it performed a coordinating function by establishing standards and expectations for the use and production of humanitarian data, and visualizations and graphics created by the community were incorporated into the HDX gallery for download and use in independent risk communication efforts. Here, we see convergence between digitally enabled coordination efforts, deeply rooted in multilateral humanitarian institutions, and the diagnostic processes that they enable. Unlike the examples above, the question of who engages with the HDX and how is not significantly pre-determined. The platform is designed to be open, and to the extent that contributors are invited and approved by OCHA, it is reasonable to expect that it is open to national authorities. To the extent that technological capacity still limits national authorities from engaging with the platform, convergence of diagnostic and coordination activities nonetheless consolidates what we can call informatic discretion outside their spheres of influence.

What is striking about these examples is not necessarily the fact that activities related to different informatic components interact; that has to some degree perhaps always been the case. What is striking is the degree to which they do so automatically, as conditioned by digital media, and within the purview of the actor driving the technological and informatic innovation. This is almost never national authorities, due to capacity issues described above. To the extent that reliance on technology and the introduction of novel actors drive informatic discretion beyond the influence of national authorities, this phenomenon is likely to be exacerbated by instances of technological and operational convergence.