The world’s 1st Crowdsourced Early Warning System for Epidemics

Creating the world’s first Crowdsourced Early Warning System for infectious disease outbreaks - just like the weather forecast, but for epidemics. The system is operated under AfyaNet - a network of National Health Institutes in Africa dedicated to leveraging digital solutions at scale for disease surveillance and forecasting.

Natalia Adler, CEO, Pebble Analytics. 

Infectious diseases account for 69% of deaths in Africa. The continent has the largest proportion of the burden of disease worldwide (21.4%) but very small proportions (0.7%) of health expenditure. 

Early identification of outbreaks often relies on traditional sentinel systems at country level. These systems often lack adequate resources and require patients to go to clinics or hospitals, which is often an arduous, expensive, and time-consuming process. People often wait until they are really sick to bring themselves in, during which time they can continue to infect others. Individuals may even choose to never go to the hospital, which means that their symptoms are never reported and the disease is able to continue to circulate, undetected. These networks also do not gather data from healthy individuals, making it difficult to pinpoint trends and demographic factors in disease that are important to forecasting. While robust models for epidemics forecast exist, they tend to operate in isolation, are ad hoc or heavily academic, and offer limited usability for application in real-world situations. 

These surveillance systems are also country-specific. There’s no standardized technological, epidemiological and policy framework for coherent surveillance and data-sharing across countries. And diseases see no borders.

Our main target are National Health Institutes (NHIs) in Sub-Saharan Africa interested in digital solutions and technology to complement their sentinel surveillance systems. After multiple webinars and consultations, we have created the AfyaNet network to respond to their needs in a cohesive and scalable manner. All partners are involved in all steps of this collaborative solution (e.g., co-developing user needs surveys, ethics protocols, privacy policies, syndromic surveys, user acquisition campaigns, analytics, etc). AfyaNet also offers a platform for knowledge exchange, where countries learn from one another and engage in disease surveillance in a standardized and scalable manner.

Infectious disease outbreaks cause economic shock waves far beyond traditional health sectors. In addition to governments and humanitarian and development organizations, we believe an early warning system for epidemics can help industry improve their business preparedness, optimize their supply chain, reduce impact on employees (e.g. from productivity to absenteeism), and anticipate impact on customers. We’re developing an Early Adopters Program with stakeholders from different industries (e.g., humanitarian, multilaterals and insurance) interested in tailored alert systems for outbreaks based on aggregated and anonymized data available as a Software-As-A-Service. NATO is a client for our outbreak alert system in Kosovo.

Complex problems beget complex solutions. Our vision is to not only create a standardized system that identifies and forecasts outbreaks early on, but to make it sustainable overtime.  

We created AfyaNet, a non-profit data oversight network that governs access and use of advanced digital tools and analytics for the identification and forecasting of outbreaks.  The innovative governance structure of AfyaNet provides an additional layer of guarantees about the creation of public goods (e.g., making outbreak alerts publicly available, reinforcing cross-country data-sharing agreements, etc), while abiding to data protection and privacy regulations (attachment). It also governs the relationship between a diverse range of collaborating public and private partners. 

To cement the creation of public goods, the GovLab will enable the consortium to generate evidence, models and methodologies with regard to: 

• How to enable a “membership good” approach to leveraging and accessing data for the creation of public goods  - including terms and conditions of possible members; 

• Ways to govern crowdsourced data in a distributed yet trusted manner;

• How to establish regional data collaboratives around crowdsourced data in a systematic, sustainable and responsible way;

• Ways to incentivize people to engage and share data;

• Approach to translate data intelligence into decision intelligence.

Citizens. Our data-gathering approach has proven to be an effective and cost-effective method for gathering information about the circulation of diseases, especially among  those who do not seek (or cannot afford) healthcare assistance when sick. We're also testing different incentive systems to provide additional value to users (e.g. a health status dashboard that can be used to track one’s health progress;  NudgeBot technology; lottery of virtual tokens, etc) and creating a level playing field to ensure access. Involving Community Health Workers, as ambassadors (with their own incentive systems), will allow us to reach the most vulnerable.

National Health Institutes (NHIs). Our solution provides them with a safe and collaborative way to exploit state-of-the-art technological solutions, enhance local capacity, and transition from reactive to predictive care. Our social impact business model provides an assurance that solutions will continue regardless of donor funding at no cost to them. Through AfyaNet, NHIs can also adopt a regional outlook of outbreaks, through a trusted framework for data-sharing, learning and contributing to a regional public good in disease surveillance and forecasting. 

Private sector. Our alerts can also help industry better forecast demand, identify supply-chain disruptions, target at-risk workers, and determine the effectiveness of crisis-intervention strategies.

Our solution is inherently global and based on collaborative solutions based on a reasonable and feasible plan that can scale due to the use of technology and standardized parameters that work for all countries (with a certain level of contextualization).

During the first year, focus will be placed on maturing our 5 pilot countries as the ‘ambassadors’ of the initiative under AfyaNet. We will solidify the governance structure of AfyaNet, especially in collaboration with Trinity Challenge partners and others (e.g., Ending Pandemic), strengthening knowledge exchange and lessons learned as a network. AfyaNet already has plans to carve out the space for horizontal cooperation. Building on this foundation, we will scale our solution to 14 additional countries (6 in year 2 and 8 in the last year). 

To increase the number of users self-reporting their health status, different UX design and incentives packages will be tested to increase, but especially retain users (attachment). We also hope to continue the collaboration with Trinity Challenge members to test what works in this area (later). We also see the opportunity to use cash transfers to compensate for the value of sharing data  and level the playing field for access among vulnerable groups.

Our proposal aims at:

• Creating the 1st standardized technological and epidemiological framework for disease surveillance and forecasting, just like the weather forecast.

• Developing technology that can be sustainable overtime that works in high- and low-income countries, with focus on context-relevant and bias-aware approaches. 

• Enabling health officials to use evidence to enact timely through a ‘decision intelligence’ dashboard and access to timely funding (via Cat Bond).

To measure progress and performance, we monitor: 

• # countries committed to digital solutions for disease surveillance and forecasting (currently 5 with LoIs, 4 soft interest).

• Preliminary foundation in place for rolling and duplicating our solution (e.g., Ethics Protocol, Privacy/Terms of Use, User Needs Survey, Communication Strategy) (currently in place for 3 countries).

• # users engaged in self-reporting their health status (currently 1,000 in testing phase).

• # of real-time alerts and notification issued (mock-up phase).

To measure impact, we created a comprehensive impact tool (soon to be open source) that shows the benefit of a 2-week forecast on outbreaks. In Kenya, for example, this advance notice could contribute to:

• 11,000 hospital days preserved

• $120,000 hospital costs saved

• $1.8m lifetime income tax preserved

• 1,434 children who keep a parent

Appropriate incentives. Our solution doesn’t need many users (i.e., in Italy, Influenza outbreaks trends can be detected with fewer than 5,000 users). However, we need loyal users committed to self-reporting their health status on a regular basis. Therefore we will conduct A/B testing to identify the optimal combination of incentive mechanisms to keep citizens (and Community Health Workers) engaged, including the most marginalized.  

Trusted governance structure. Countries are naturally skeptical about sharing data. Therefore, establishing a framework for regional data-sharing requires trust, which in turn can be achieved through tco-creation and a clear value proposition that works for all involved. We will reach this sweet spot by further consolidating AfyaNet. 

Evidence to action. Our collective experience shows that having just the best data is not enough to drive (timely) actions. We will focus on translating ‘data intelligence’ into ‘decision-intelligence,’ while also injecting timely access to funding through the creation of parametric risk-transfer instruments that link data-drive triggers with prompt insurance payouts. The assumption is that having access to funding early enough will not only expedite the decision-making process but will also shift decisions towards less costly preventive measures and strengthening resilience.

• Center of Design, Northeastern University, USA

• The GovLab, New York University, USA

• Mitiga Solutions, Spain

• National Institute of Communicable Diseases, SA

• Centro Nacional de Saúde and Eduardo Mondlane University, Mozambique

• Center for Virus Research, Kenya

• Institut National d'Hygiène Publique, Côte d’Ivoire

• Institute for Scientific Interchange Foundation, Italy 

• Barcelona Supercomputing Center

Stopping the next pandemics is a complex endeavor and it requires an integrated response that combines technology, data science, design and governance/policy structures, along with market solutions to ensure sustainability of the initial investment. We have seen great ideas fail (or dwindle) because they were either one dimensional (focusing only on technology) or donor-dependent with no sustainable pathway. Without connecting the dots across different disciplines and engaging in collaborative solutions, the mission of the Trinity Challenge may be hampered by the persisting ‘cycle of panic and neglect’ that characterizes pandemic response. 

We believe our solution combines all elements required to move the needle forward in a sustainable fashion, while creating the foundation to scale and expanding collaborative solutions. The implementation of cost-effective technological solutions - under a strong infrastructure - can help low-income countries leapfrog data constraints and substantially enhance local capacity for risk management of emerging and re-emerging diseases, to transit from reactive to predictive care. 

Our consortium is equipped to respond to the barriers identified (e.g. appropriate incentives, trusted governance structure, driving evidence to action) by bringing together technology, design, policy, and governance in a balanced way to deliver sustainable public goods that prioritize prevention and resilience building.

Pebble Analytics and Mitiga Solutions are part of Microsoft’s Startup Program, so we would like to continue exploring synergies with them, particularly around cloud infrastructure. 

In addition, we envision collaborations with:

• Bill & Melinda Gates Foundation: to scale AfyaNet through existing entry points with NHIs and leverage their expertise to expand outbreak monitoring for diseases such as malaria. 

• The Behavioral Insights Team: to amplify and help us test our incentive packages in multicultural settings.

• John Hopkins Bloomberg School of Public Health: to leverage their dashboard expertise, particularly around setting up thresholds for alert notifications.

• Infosys: to leverage their expertise to create seemingless digital experiences to attract and retain users in different settings.