Crowdsourcing Real-Time Viral Disease and Pest Information: A Case Study of Nation-Wide Cassava Disease Surveillance in Uganda

These attacks can have a devastating impact on smallholder farmers, who depend on these systems for their livelihoods. A key component of any proposed solution is a robust disease surveillance network. However, current surveillance efforts are limited by a lack of resources, both human and financial. Crowdsourcing with farmer crowds that have access to mobile phones offers a viable option to provide real-time surveillance data on viral disease and pest incidence and severity. In Uganda, where viral disease attacks on crops are a leading cause of food insecurity and poverty, having a functional surveillance system is critical. The cassava crop, an important food security crop for Sub-Saharan Africa and other regions, is particularly vulnerable to pests and viral diseases. Cassava is the second most important food and security crop in Sub-Saharan Africa, especially amongst smallholder farmers, because it can easily be grown in poor soils and requires few inputs. Although cassava is known to survive under harsh conditions, its yield is greatly affected by pests and viral diseases. The current surveillance system in Uganda relies on expert surveys, which are limited in their ability to provide real-time actionable surveillance data. These surveys are often delayed or limited in scope due to budget constraints. Citizen science and crowdsourcing have been shown to be effective in supplementing long-term focused systems, such as the surveillance system employed in Uganda. Crowdsourcing can provide ad hoc data input from citizens or crowds nearer to the phenomenon of interest, making it a viable option for real-time surveillance data collection. Mobile telephony has made it possible to leverage crowdsourcing for this type of problem, with several example implementations already existing. In the Ugandan context, smallholder farmers and extension service workers can provide supplemental surveillance data more frequently and more efficiently using smartphones. A mobile ad hoc surveillance system, AdSurv, was implemented to enable farmers, extension workers, and agricultural experts to provide near real-time geo-tagged surveillance data for monitoring cassava crop health across Uganda. The AdSurv system involves the crowdsourcer broadcasting a surveillance task of a subject of interest to the agents, who then send in images and text data pertaining to the request. The system uses a Knowledge Collection from Volunteer Contributors (KCVC) methodology, which compensates collectors with mobile phone data credit, micro payments, or other types of incentives. A pilot study was conducted in several regions of Uganda, spanning 76 weeks, to test the utility of crowdsourced surveillance data in informing action. The pilot study provided evidence of the effectiveness of crowdsourcing in real-time surveillance data collection and informed the development of algorithms to automatically classify disease.