Lectures

The hype and hope of data for healthcare in Africa

Promotional materials from the global campaign to achieve Universal Health Coverage by the year 2030. Copyright UHC2030 – reproduced here under ‘fair use’ for academic purposes.

“Health for All?” critically explores global moves towards Universal Health Coverage and its language of rights to health, equity, social justice and the public good. Highlighting emerging ethnographic and historical research by both young and established scholars, the series explores the translations and frictions surrounding aspirations for “health for all” as they move across the globe. The series is edited by Ruth Prince.

Figure 1. A data scientist examines the quality of a smartphone photo of a blood sample infected with plasmodium falciparum (all images by author unless otherwise noted)

In November of last year, rumours spread in Tanzania that Ebola had reached the country for the first time in its history. The World Health Organization (WHO) accused the government of suppressing positive test results, and the government subsequently scolded the WHO for its unsubstantiated accusations. Nobody knew what was to come only a few months later.  

This piece was written shortly before I left Tanzania in March this year. I had been living and working there for two years, and by the time I left, COVID-19 was becoming a part of everyday conversations. Some of my interlocutors and their colleagues in the data and computer science world were also being affected; for instance, one was unable to return to studies in China while another cancelled their attendance at a family event after he had travelled abroad.

Now back in Norway, I follow the news online and keep in touch, through WhatsApp, with my friends and interlocutors in Tanzania. Quite quickly after the first case was confirmed, the government put in place a range of measures which have included quarantining foreign visitors as well as testing and contact tracing of suspected cases. They are also discussing how new forms of digital surveillance, currently being deployed across the world – including in Norway – may also be designed and rolled out.[1]

My research over the last two years has centred on the phenomena of data in Tanzania. Most recently, it has involved spending time with Tanzanian scientists and entrepreneurs as they go about designing and testing data-driven health innovations they believe will help to provide health for all. Some of those who I have got to know have begun to use their expertise to help their country and their fellow citizens tackle the COVID-19 crisis. This is a fast-moving world and it is impossible to know what will emerge from their involvement. What follows may provide some helpful contextualisation around how locally developed, data-driven solutions could unfold over the coming months. However, my focus in this piece is on the role data-driven technology has been playing in Tanzania pre-COVID-19.

As global momentum builds behind the Universal Health Coverage (UHC) agenda, new thinking is taking place around how this drive might best be directed. Some of this thinking emerges from a techno-optimist discourse that draws from and nourishes the rapid expansion of digital and data infrastructures across the world. In sub-Saharan Africa, this discourse has generated a sense of possibility that technologies, including artificial intelligence (AI), robotics, and drones, could improve healthcare quality and access for the rural and urban poor. As the headline of one opinion piece in Newsweek by a Nigerian entrepreneur, Adebayo Alonge, put it, “How AI can help Africa get universal health care before America’ (2017).

These claims are part of a technological movement that has swept the world, generating both hope and counterclaims of hype. Technologists, activists, and politicians of all stripes have begun invoking the phrase “the fourth industrial revolution”, coined five years ago by the economist and founder of the World Economic Forum, Klaus Schwab. Across Africa, many governments, fearful of losing an opportunity and falling behind, have launched task forces to investigate the possibilities and dangers of AI and other emerging digital technologies.

At the global level, large public health institutions are funding ambitious programmes. At the recent UN high-level meeting on UHC in September, the Rockefeller Foundation launched a $100 million programme, Precision Public Health, that will use AI machine learning to improve public health in the global South. It aims to put predictive analytic tools in the hands of primary and community health workers in order to address poor maternal and child health outcomes. Furthermore, before the current COVID-19 crisis, one of the world’s largest AI conferences, the Eighth International Conference on Learning Representations (ICLR) was going to be held in Africa for the first time, in Ethiopia, and was expected to attract at least 7,000 participants. In such moves, a worldwide, Silicon Valley-inspired, capitalistic-centred tech movement connects to the continent of Africa and its varied histories, with promises of “leapfrogging” – surpassing the rest of the world in healthcare innovation.

In the context of healthcare, leapfrogging often means imagining new pathways that utilise technologies and automation to skip over traditional steps, for instance, those that rely heavily on human expertise. In Tanzania, a new generation of scientists and technologists are joining this global tech movement. In their homes, and in the university campuses and tech hubs of cities such as Dar es Salaam, Dodoma and Arusha, these young Tanzanians are teaching themselves and each other data science through open software, data and learning materials accessible online. They are taking Massive Open Online Courses (MOOCs), sharing books and YouTube videos over WhatsApp, competing in online competitions like Kaggle, and meeting together physically in coding clubs. Where possible, they attend regional and international conferences and workshops where celebrities from the data science world often speak on the theme of data science for the ‘social good’. Others, with advanced degrees, including PhDs in artificial intelligence, have also begun returning from overseas, in a move that has been dubbed a ‘brain gain’.[2]

Many of these Tanzanian scientists and technologists are thinking about how they can deploy their new skills in data science to help provide health for all. One key challenge they invariably identify is the shortage of well-trained health workers; a shortage which they see as only growing. In labs and health clinics, they have begun experimenting to find out how they can make up for this human resource gap. Some are using smartphones and training deep learning models to speed up the process of in vitro diagnostics and supplement the human expertise that is required. Others are using drones and computer vision learning to map out malaria breeding sites. Others still are developing AI-powered algorithmic health assistants to support both the general public and healthcare workers in their efforts to prevent and even diagnose health problems, from diabetes to cancer to mental illness.

Their efforts enter a context in which the lay public and healthcare workers have been, over the past decade, turning to their own mobile devices to access health expertise. During a recent conversation with Said, an old friend who owns a garage in northern Tanzania, he reached for his smartphone and scrolled through the gallery. Eventually he found what he was looking for: a series of photos he had taken of the inside of his throat. He explained that he had WhatsApped them to one of his customers, a doctor.[3] After a serious road accident the year before, Said no longer had health insurance and going to the hospital was too expensive. This way he could get medical advice through the relationships he had built up around his livelihood.

It is not only Said who is accessing health information through his phone. It is estimated that between a quarter and a half of people in countries such as the UK and US diagnose through the internet, but there is evidence that this is also happening in East Africa, where trained healthcare workers are even more out of reach. Pharmacists and the lay public, and even madaktari wa mtaani (street doctors) in both urban and rural areas are accessing both English and Kiswahili language medical resources. For instance, one of the top five most downloaded medical apps in the Google Play Store in Tanzania is the Kiswahili-language tiba (meaning “medical treatment”), which offers information on maradhi (sickness), the health benefits of mimea (plants), as well as advice on uchawi (witchcraft). A clinical officer in an urban health centre in northern Tanzania, who has begun seeing a growing number of people who have consulted the internet for medical advice before seeing him, has taken to referring to these patients as wagonjwa dot com (“dot com patients”).

Figure 2 A notepad from Deep Learning Indaba, Africa’s foremost AI gathering, next to a box of blood slides in a lab.

Returning to these young Tanzanians, it is clear there is an excitement among them about the potential for data analytics for their own country. One young male participant on a WhatsApp group I am a member of recently wrote enthusiastically about the possibilities for Africans to “ride the wave” of AI, arguing that apart from witchcraft, Africans had been left behind in almost everything.

Invoking witchcraft, the young man pushed back against the narrative of Tanzania as a country characterised by anti-science superstition, a narrative he thinks portrays his country as backward and outdated (for a related example about the young American scientists of the 1960s, see Kaiser & McCray 2016). Instead, this man, like other young Tanzanian scientists, sees an opportunity for himself and his country to be at the forefront of science. These young Tanzanians are also pushing back against the historical narrative that pre-destines sub-Saharan Africa as a producer of raw products – in this case, “raw” data (Gitelman 2013). Young Tanzanian data enthusiasts have begun to explore how data can be collected, cleaned, refined and utilised within their own country to help solve problems.

Figure 3. A tablet in a rural health dispensary, loaded with an AI-powered app for the screening of NCDs

These more advanced data-driven solutions are not the first digital health technologies to make their way into rural East Africa. The first attempt at telemedicine in Tanzania, using very high frequency (VHF) radio, was in the early 1990s. But the recent proliferation of low-cost mobile phones and tablets has created significantly more opportunities. For instance, in Tanzania the donor-supported government is currently scaling-up their digital immunization registry system, which has brought computer tablets to rural health centres and dispensaries. The data produced is already being analysed by healthcare workers, the government, as well as by a North American AI company who is seeking to optimise the supply chain and predict stock-outs. Such foreign data-focused companies, who are invariably better resourced, have also begun operating and experimenting across Africa. The health AI and telemedicine company, Babylon Health — well known in the UK because of its growing presence within the National Health Service — has been operating in Rwanda for the last two years. After a $500 million investment round, raising their valuation to over $2 billion, they are set to further expand their operations across the continent.

One story that can be told would explore how Africa has become an attractive market for these foreign data-focussed companies interested in the health sector. Aside from Babylon, giant corporations such as Google and IBM have already established bases in Ghana and Kenya, respectively. In the current COVID-19 crisis, there is little doubt that these companies will incorporate Africa into their remit of activities, surveillance or otherwise. Some might argue that the tale of these companies in Africa is the most important story to tell, and to critique. However, as important as this might be, a singular focus upon them would ignore how science, technology and innovation emerges from the continent and its people, rather than simply happens to them (Mavhunga 2017).

In his recent book Deep Medicine, the American medical doctor and futurist, Eric Topol, used an example of a small AI start-up in the US which was able to develop a better digital device to detect strokes than Apple, stating that “in the era of AI medicine, David can definitely still beat Goliath” (Topol 2019). Across Africa, there are plentiful Davids: young, highly educated East Africans with advanced degrees who are establishing their own AI medical projects and start-ups. They argue forcefully, and persuasively, that they are far better situated, if not necessarily resourced, to be in control of the design of health technologies for their own country.

There is little doubt that digital and data-driven technologies have now become an established feature of both healthcare generally and the UHC agenda more specifically. A recent survey in Tanzania among technologists at one university showed that the majority of respondents saw these technologies as having the most benefit in the health sector; a finding which is reflected in the increasing number of new data-driven health projects in the country.[4] Their increasing presence along with their potential for both huge gains in health outcomes and huge dangers means they must be taken seriously, sympathetically, and critically (see for example, Erikson 2018). The automation that is expected to be achieved through these technologies is, as social scientists are well aware, not simply a technical issue, but a deeply moral, cultural, and social one (for an early example of this in AI see Forsythe 1993). In health futures that view biomedicine as no longer exclusively in the hands of the doctor, it is important to pay attention to these dimensions among a growing collection of new actors.

In Tanzania, these new non-medical actors, who do not themselves labour intimately and desperately over fragile and ailing bodies, rarely invoke a “heart for the work” in the same way as the Malawian medical students, described by Claire Wendland (2010). However, they are no less shaped –and driven — by moral commitments and values around what it means to be a good and responsible data scientist working on issues in the healthcare sector.

As these young scientists and entrepreneurs go about creating radical new health futures through the design and building of automated and computerised ‘healthcare assistants’ they bake into these assistants and their algorithmic brains (and hearts?) an assemblage of moral values, cultural norms and social relationships.

A detailed focus on the data science and health community in Tanzania allows us to understand how these dimensions come together as actors grapple with the unruly, uncertain and stubborn Tanzanian bureaucratic, healthcare and data worlds and their histories. Understanding how this happens is an important step in getting to grips with the optimistic, even quasi-utopian, talk of digital and data-driven technologies (see also Ruth Prince’s piece on digital registration), leapfrogging, and Universal Health Coverage.

Figure 4. Anti-bacterial hand sanitizer installed in a bajaj (auto rickshaw) by a Tanzanian artificial intelligence start-up. Photo taken by the co-founder of this startup.

Finally, we might reasonably ask how all of this could relate to what is happening in this current moment of COVID-19. The simple answer is that we don’t know. Through WhatsApp, I am regularly introduced to new AI symptom assessment checkers and receive stories about the new forms of digital surveillance that are being designed to help tackle the crisis. My interlocutors are similarly trying to work out how they might contribute their data and computer science skills in the current moment. One company I have been spending time with has already begun to lend their data science expertise to the Tanzanian government.

One Tanzanian AI company is also doing something rather unexpected. In a Telegram group, the founder of that company posted this: “Hey guys, I read something a few days ago. ‘For all you Machine Learning and AI people out there looking for a way to help on the COVID-19 crisis, please do so by washing your hands multiple times per day with soap and water as well as avoiding crowds and social distancing’. Found it funny and thought I’d share!”. A few days later, putting his money where his mouth was, his company installed, without charge, antibacterial hand sanitizer dispensers in some bajajs [petrol rickshaws] used by many Dar Es Salaam residents. While critical voices will, often quite legitimately, question the rush to develop digital tech solutions to the current crisis, it would be a shame to pigeonhole technologists by glossing over the nuance and complexities of their work and their lives.

Notes

[1] For instance, in Norway the largest mobile operator, Telenor, is working with the government to track cases https://www.bbc.com/news/av/technology-52236559/coronavirus-mobile-data-helps-norway-track-cases

[2] https://qz.com/africa/1128778/africa-brain-drain-to-brain-gain-african-elite-graduates-head-home-as-brexit-trump-eu-close-doors/.

[3] In Tanzania, the term doctor (daktari in Kiswahili) is used to refer to many healthcare professionals without a medical degree, such as clinical officers.

[4] https://saharaventures.com/publication/Artificial-Intelligence-In-Tanzania-Whats-Happening

Works Cited

Alonge, A. 2017. How AI could revolutionize healthcare in Africa long before the United States. Newsweek, 30 October (available on-line: https://www.newsweek.com/artificial-intelligence-us-healthcare-africa-693849, accessed 23 November 2019).

Erikson, S. L. 2018. Cell Phones ≠ Self and Other Problems with Big Data Detection and Containment during Epidemics: Problems with Big Data Detection and Containment. Medical Anthropology Quarterly 32, 315–339.

Forsythe, D. E. 1993. Engineering Knowledge: The Construction of Knowledge in Artificial Intelligence. Social Studies of Science 23, 445–477.

Gitelman, L. (ed) 2013. ‘Raw Data’ is an Oxymoron. Cambridge, Massachusetts ; London, England: MIT Press.

Kaiser, D. & W. P. McCray (eds) 2016. Groovy Science: Knowledge, Innovation, and American Counterculture. Chicago: University of Chicago Press.

Mavhunga, C. C. (ed) 2017. What Do Science, Technology, and Innovation Mean from Africa? Cambridge, MA: MIT Press.

Topol, E. 2019. Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. New York: Basic Books.

Ubwani, Z. 2020. Tanzania: It’s Businesses As Usual At Namanga Despite COVID-19. The Citizen (available on-line: https://allafrica.com/stories/202004050079.html, accessed 8 April 2020).

Wendland, C. L. L. 2010. A Heart for the Work: Journeys through an African Medical School. Chicago ; London: University of Chicago Press.


Tom Neumark is a Postdoctoral Fellow in Social Anthropology at the Centre for Development and the Environment, and the Institute of Health and Society at the University of Oslo. Tom has a long-standing interest in East Africa, and in the ethical, political and economic dimensions of development, humanitarian and global health interventions. His current research, concerning data-driven technologies and innovation in health in East Africa, is part of the research project, Universal Health Coverage and the Public Good in Africa, led by Ruth Prince and funded by the European Research Council.


Leave a Reply

Your email address will not be published. Required fields are marked *