Surveillance as a side effect of service
Completed
Crop pests and diseases present a significant threat to the livelihoods of smallholder farmers in Ghana. There are two
key barriers to addressing crop losses due to pest and diseases; the lack of access to agricultural extension advice, and
the lack of data to estimate local prevalence of important pests and diseases. This brief project description includes the objectives,
methods, results and conclusion of the project conducted using the GCE Phase I award.
The aim of the project was to develop an innovative and transformative solution for surveillance of crop pests and
diseases. After finalisation of the project design, the specific objectives were:
- To design and implement a system in which farmers report signs of crop pests or diseases, and receive immediate
feedback including a presumptive diagnosis and management advice. - To estimate area-level prevalence of priority crop pests and diseases based on farmer reports.
- To assess the acceptability, accessibility and uptake of this crop pest and disease reporting system.
Conclusion
The core idea of the project – ‘surveillance as a side effect of service’ – was realised. A real-time pest and disease diagnosis and management tool accessible from a feature phone has been implemented, and the resulting data can be used to construct maps of priority pests and diseases of two major food crops in the Ashanti region in Ghana. Despite some challenges in using the system, the service was perceived highly favourably, with farmers reporting that it helped them to address pests and diseases on their farm, and reduced their costs of travelling to the district agriculture office for advice. An innovative and scalable statistical method for deriving crop pest and disease prevalence estimates using small area estimation methods was developed. Though too few data points were available to generate reliable estimates, proof of principle has been achieved. Future work will require adding a standalone user registration to enable full use of reported data, and validating the presumptive diagnoses. The system is readily extendable to other settings and crops, and is particularly suited to settings with low literacy levels and low ownership of smart phones.
References
[1] Rao JNK, Molina I. Empirical Best Linear Unbiased Prediction (EBLUP): Theory. Small Area Estimation, John Wiley & Sons, Ltd; 2015, p. 97–122.
[2] Kreutzmann A-K, Pannier S, Rojas-Perilla N, Schmid T, Templ M, Tzavidis N. The R package emdi for estimating and mapping regionally disaggregated indicators. Journal of Statistical Software 2019;91.
[3] Baddeley A, Rubak E, Turner R. Spatial point patterns: methodology and applications with R. CRC press; 2015.
[4] Rojas‐Perilla N, Pannier S, Schmid T, Tzavidis N. Data‐driven transformations in small area estimation. Journal of the Royal Statistical Society: Series A (Statistics in Society) 2020;183:121–48.
[5] Rogers EM. Diffusion of innovations. 5th Edition, Simon and Schuster; 2010.
[6] FAO. FAO GIEWS Country Brief on Ghana. http://www.fao.org/giews/countrybrief/country.jsp?code=GHA
Challenges
- Challenge 1: Identifying a suitable implementation partner and country after original proposed partnership (WeFarm in Kenya) did not advance. Response: A new partnership was established with Farmerline in Ghana, which involved in-person meetings (Seattle and Brussels) and teleconferences to discuss the project aims, a visit to Ghana to define project activities, and agreement to collaborate (subcontract signed 16 December 2019).
- Challenge 2: Low literacy levels did not support an SMS-based reporting system as originally proposed. Response: Toll-free USSD code used to initiate IVR call, which still allowed the system to function using basic feature phones as per the original proposal. This modestly increased costs compared to a SMS-based system, as audio recording of the survey and responses was required.
- Challenge 3: COVID-19 lockdown delayed roll-out. Response: Once lockdown measures were eased, community workshops were held, with smaller group sizes, and masks, hand sanitiser and social distancing in place.
- Challenge 4: COVID-19 lockdown prevented the planned collaboration with a local university partner to conduct an agricultural survey to validate the presumptive diagnoses. Response: There was no viable alternative to incorporate an agricultural survey within the budgetary constraints. This will be proposed for Phase II.
- Challenge 5: Didn’t reach enough data points to send early warnings or community-wide alerts. Response: Phase II proposal will include a power analysis to estimate number of reports per community/crop/pest required to generate statistically robust estimates, at which point alerts can be sent.
