Modeling Early Risk Indicators to Anticipate Malnutrition (MERIAM)
Project duration:
2017-2021
Project Consortium:
Action Against Hunger (US)
University of Maryland (US)
Graduate Institute Geneva (CH)
Johns Hopkins University (US)
Website:
https://www.actionagainsthunger.org/meriam
Project Description:
The project was funded through a grant of the Department for International Development of the United Kingdom (DFID). Its central aim was to identify, test and scale up cost-effective means to improve the prediction and monitoring of undernutrition in difficult contexts, in such a way that it enables an effective response to manage and mitigate nutritional risk. The cross-disciplinary research team that formed the MERIAM consortium was comprised of experts from several pertinent fields and drew from organizations and institutions that are world leaders in nutrition-related practice and research.
The project specifically aimed to produce techniques and tools that are suited to strengthen early warning systems by both forecasting an increased risk of undernutrition and identifying the key drivers of that risk, and, generating scenarios that demonstrate how the timing and type of services provided may affect the impact of a shock on communities, including consideration of the cost-implications of the response for the outcome achieved. The project therefore concentrated on the early (‘leading’) indicators of undernutrition, leveraging a wide variety of existing and accessible data to rigorously capture its causal factors and dynamically model its fluctuation in contexts where this information is most urgently required.
Complementary types of quantitative methodologies wer employed to take account of the data landscape and to appropriately reflect the intricacies of potential contributing mechanisms. First, spatio-temporal econometrics were used to identify empirical relationships, paying close attention to geographic (e.g., local conditions) and time-series properties (e.g., seasonality), endogeneity (i.e., reciprocal causality), and shocks (e.g., conflict, drought, spikes in food prices). Second, computational modeling was used to explore undernutrition as an emergent phenomenon and to trace explanations for its sources. Resulting simulations provided a versatile means for analysis of “what-if” counterfactuals relevant to policy and practice (e.g. undernutrition fluctuation in response to shocks, seasonal patterns and variations, early vs. late humanitarian response, etc.) and for linking micro-level attributes and behaviours, meso-level context, and emergent macro-level outcomes.
The geographical focus of the project was on Africa—Kenya, Somalia, Uganda and Niger, in particular. This allowed to fully harness both the available data for these country contexts as well as existing operations of Action Against Hunger in those countries. With the project focused primarily on the process and applications of forecasting rather than on concrete findings of risk, it could though derive insights relevant for malnutrition early warning valid beyond the specific country contexts studied.