How modern technologies can help to achieve “health for all”

So-called “frontier technologies” offer a transformative potential towards reaching universal health coverage in low and middle income countries. This will only be achieved, though, if the technology-based solutions are scaled-up and systematically integrated into existing healthcare structures. Moreover, fundamental technical, ethical and legal questions regarding data security, fairness, non-discrimination and privacy need to be addressed. The World Health Organization (WHO) could assume a leading role in facilitating the development of a global regulatory framework.

Fondation Botnar contributes to a partnership between D-tree International and the Zanzibar Ministry of Health to support the roll out of a national digital community health system to improve maternal, newborn, and child health services in Zanzibar. Photo: © D-tree International

Despite much progress being made in global health, appalling inequalities in healthcare still exist. At least half of the world’s population lack access to essential health services, and even with efforts made in recent years, one third of the global population will remain underserved by 2030, unless groundbreaking advances are achieved (UHC Political Declaration, 2019). Digital technologies have been advertised as the saving grace in avoiding stagnation. However, their potential for healthcare is still to be clearly defined and a number of ethical, legal, and societal questions need to be answered.

It is in this light that Switzerland’s decision to revise its Health Foreign Policy in May 2019 and to underline the importance of digitalization and modern technologies for healthcare, is to be welcomed (BAG: Swiss Health Foreign Policy 2019). The revised policy stresses that new technologies need to be leveraged in order to improve healthcare. Likewise, numerous other agencies and commissions including the World Health Organization (WHO) and the UN High-level Panel on Digital Cooperation emphasize the potential offered by so-called frontier technologies. In a similar vein, they warn that if we want to achieve universal health coverage (UHC) by 2030, digital technologies and artificial intelligence (AI) need to be leveraged not just in the global north, but even more so in low and middle income countries (LMICs) (WHO: Draft global strategy on digital health 2020-2024) (UN High-Level Panel on Digital Cooperation). They stress that it is imperative that these technologies are fully integrated into existing healthcare structures.    

The potential of digital health for LMICs

As a foundation, we want to use our position and resources to catalyze the potential of digital health in LMICs. An example from Tanzania illustrates what such an integrated approach can look like. In 2019, together with several implementation partners, we launched a proof of concept initiative called “Afya-Tek”, aimed at integrating digital technologies into a new responsive people-centered health system in Kibaha, Tanzania (Fondation Botnar and partners launch Afya-Tek). Currently, the Tanzanian health system is highly fragmented, severely limiting both quality and access to medical treatments. “Afya-Tek” is harnessing digital technology to better connect all the health system actors, including using predictive analytics and biometric identification to improve medical records and decision-making. If successful, this initiative will demonstrate how technology can overcome one of the most damaging weaknesses of health systems – fragmentation.

Another example is the DYNAMIC project, a partnership we launched this year with researchers from Unisanté. Every year, 3.3 million children die from an acute febrile episode, and in Tanzania the ever-increasing use of antibiotics is resulting in bacterial resistance and ineffectiveness (Fondation Botnar announces a new partnership with Unisanté). Studies show that nine times out of ten, an antibiotic prescription is not necessary. To combat flawed practices, the project is integrating dynamic clinical algorithms into frontline healthcare to guide and train health workers. Such technological applications have the potential to improve treatment and to drastically curb unnecessary prescriptions of antibiotics.

Afya-Tek team visit a private drug dispensary in Kibaha, Tanzania. Afya-Tek is a digitally-enabled approach that links community health workers, health facilities and private drug dispensers to better support patients. Photo: © D-Tree International

Limitations, problems and challenges

Through moving away from the hype, and focusing on practical application, digital technologies can fundamentally transform healthcare in LMICs. Yet, while these examples clearly illustrate the potential benefits of integrating digital health technology and AI into healthcare systems, there is a strong need to frankly assess how such integration can be successful and which steps we need to pursue in order to reach this goal.

Generally, and perhaps most fundamentally, we need to concede that there is not always a technical solution to a sociopolitical problem. The current state of global health (and of national health systems) is to a large degree a consequence of specific, often historically grown, political and economic factors. Moreover, health systems are institutionally complex. A local or national health system that is crippled by severe underfunding, economic austerity or structural injustice will not be simply turned around through implementing new technologies. Digital technologies are no silver bullet solution.

Relatedly, previous experience shows that it is not sufficient to merely transfer technological innovations from high-income countries and deploy them to LMICs. Local requirements, existing infrastructure conditions and socio-economic possibilities in resource-poor settings need to be taken into account more systematically. As has been observed by Hausmann-Muela and Eckl in a different context: “Far too little attention is paid to implementation challenges and conflicts between local needs and universal solutions. Accessibility, acceptability, and integration of tools into local health care systems all depend on the support of local populations and experienced public health staff, yet their perspectives are commonly overlooked when health campaigns are planned” (Hausmann-Muela and Eckl, 2015). Here the global health community and especially funding and donor organizations need to self-critically reflect upon longstanding practices: rather than initiating ever more vertical, top-down projects and pilots, we need to focus on scaled-up, systemic approaches that are tailored to existing infrastructure and to the needs of local populations and healthcare workers. This can best be achieved through mutual learning and through the involvement of and co-creation with local stakeholders.   

Other challenges pertain to the technological requirements: For example, it is important to address the massive digital divide in internet connectivity that still prevails in many LMICs. A recent report suggested that in order to achieve universal, affordable, and good quality broadband internet access in Africa by 2030, “an estimated additional $100 billion would be needed to reach this goal over the next decade” (Broadband Commission for Sustainable Development, p.16, 2019).This  assessment needs to be taken into account in the design, development, and implementation of solutions – for instance by making sure that they also work in offline settings.

Furthermore, tech developers need to guarantee that the algorithms embodying the core of artificial intelligence and machine learning are based on representative, fair and non-biased data sets. All too often, the data privileges white, male probands from the global north, while little attention is paid to ethnic, geographic, cultural or gender diversity. “Biases in the data often reflect deep and hidden imbalances in institutional infrastructures and social power relations”, the authors of a recent Nature article claimed (James Zou and Londa Schiebinger, 2018). They describe the case of a diagnostic software for skin-cancer that was based on more than 120.000 photographs, mostly from Google Images. It turned out, however, that the software performed poorly on non-white skinned people. The reason: Researchers had fed the algorithm predominantly with images of white skinned people - fewer than 5% of these came from dark-skinned individuals. The result was severe performance issues with the software favoring white-skinned people in accurate analysis (James Zou and Londa Schiebinger, 2018).

A similar problem persists with regard to data from LMICs: Substantial efforts are needed to improve the comprehensive collection and quality of health data in resource poor settings. Better and more granular electronic health data will not only allow public health agencies to better monitor population health and detect disease outbreaks, to target interventions and to allocate resources more efficiently. It will also enable the construction of locally useful applications. The adoption of standardised medical terminologies and the establishment of local data dictionaries would greatly facilitate such developments (Willem G. van Panhuis et al., 2014).

In Zanzibar, D-tree is supporting the Zanzibar Ministry of Health to roll out national digitally-enabled community health worker program. Photo: © D-tree International

Above all, we need to ensure that the needs and interests of vulnerable populations – e.g. women, minorities, children and adolescents – are better represented in the development of new technologies, but also that they are better protected against discrimination and mistreatment through regulatory and policy frameworks. Data ethics do not only pertain to fair and unbiased datasets, but also data ownership and data protection - a key area where governance is needed at a local, regional, and global level.

In order to take advantage of frontier technologies, these key issues must be addressed. Technology must not be a standalone solution, but instead be linked into every level of the health system to ensure it facilitates and includes access to treatment.

Policy implications

In light of the opportunities, as well as the fundamental challenges, there is a clear need for political and regulatory guidelines, norms and standards on a global level. This is particularly true for small and resource-poor states who may lack the administrative capacity to develop their own national regulatory and policy frameworks, thus not allowing them to keep up with the speed at which technological developments take place.

Many of the challenges mentioned result from a lack of governance in data collection, storage, and ownership. A global architecture could be one step forward, ensuring that these developments help improve the health and wellbeing of all citizens, especially marginalised groups. With this in place, it would preferably lead to the evolution of a regulated health data ecosystem, guaranteeing information security and the right to privacy, while allowing for comprehensive data collection, analysis, and sharing. Ultimately, health data might become a global public good that can be shared and used to create value for society (Patty Kostkov et al., 2016). 

It is therefore important to look to organizations such as the UN with its High-level Panel on Digital Cooperation, the World Health Organisation, the European Union or governments like Switzerland for policy initiatives that help to balance digitalization in healthcare with key requirements such as accountability, quality, security and access. Ultimately, a global regulatory framework is needed that aligns and protects the needs of all stakeholders – from individuals, to governments, to research institutes, to private sector companies and other organizations (Effy Vayena et al., 2018).

Conclusion

If we want to reach the health goals set forth by the UN Sustainable Development Agenda, we need to be bolder in what we are currently doing. There is great potential in digital technology driving us forward, but only if several challenges are addressed: we need to start now by co-creating inclusive and systemic interventions, adapting solutions to strengthen existing health systems, improving datasets and algorithms, and creating clear regulatory policies for health data. If we manage to tackle these ethical, legal, and societal issues, digital technology and AI can help us move towards achieving universal health coverage.

References

• UHC Political Declaration, 2019, https://www.un.org/pga/73/wp-content/uploads/sites/53/2019/05/UHC-Political-Declaration-zero-draft.pdf
• Bundesamt für Gesundheit (BAG): Swiss Health Foreign Policy 2019, https://www.bag.admin.ch/dam/bag/en/dokumente/int/GesundheitsaussenpolitikderSchweiz2019%E2%80%932024.pdf.download.pdf/GesundheitsaussenpolitikderSchweiz2019%E2%80%932024.pdf
• World Health Organization (WHO): Draft global strategy on digital health 2020-2024, https://www.who.int/docs/default-source/documents/gs4dh.pdf?sfvrsn=cd577e23_2; UN Secretary General’s High-Level Panel on Digital Cooperation: The Age of Digital Interdependence, https://digitalcooperation.org/wp-content/uploads/2019/06/DigitalCooperation-report-web-FINAL-1.pdf
• Fondation Botnar: “Fondation Botnar and partners launch Afya-Tek”, https://www.fondationbotnar.org/fondation-botnar-and-partners-launch-afya-tek/
• Fondation Botnar: “Fondation Botnar announces a new partnership with Unisanté”, https://www.fondationbotnar.org/fondation-botnar-announces-a-new-partnership-with-unisante/
• Hausmann-Muela and Eckl: “Re-imagining malaria – a platform for reflections to widen horizons in malaria control”, Malaria Journal 14 (2015), https://malariajournal.biomedcentral.com/articles/10.1186/s12936-015-0703-6
• Broadband Commission for Sustainable Development: “Connecting Africa Through Broadband”, October 2019, https://www.broadbandcommission.org/Documents/working-groups/DigitalMoonshotforAfrica_Report.pdf, p.16
• James Zou and Londa Schiebinger: “AI can be sexist and racist — it’s time to make it fair”, Nature, 18 July 2018, https://www.nature.com/articles/d41586-018-05707-8
• Willem G. van Panhuis et al: “A systematic review of barriers to data sharing in public health”, BMC Public Health, 14:1144 (2014), https://bmcpublichealth.biomedcentral.com/articles/10.1186/1471-2458-14-1144
• Patty Kostkov et al.: “Who Owns the Data? Open Data for Healthcare”, Frontiers in Public Health, 17 February 2016, https://www.frontiersin.org/articles/10.3389/fpubh.2016.00007/full
• Effy Vayena et al.: “Policy implications of big data in the health sector”, Bulletin of the World Health Organization, 96 (2018), https://www.who.int/bulletin/volumes/96/1/17-197426.pdf