AI Model Predicts Food Insecurity in Sub-Saharan Africa by Integrating Climate Data

New research introduces a novel approach to predicting food insecurity across sub-Saharan Africa by integrating climate variability data with socioeconomic factors and market disruptions. The study highlights that broad climate patterns, such as El Niño, which are typically tracked months in advance, are not yet systematically incorporated into early-warning systems for food security.

To address this gap, researchers developed "sensitivity regimes" that categorize regions based on how their vegetation responds to the El Niño Southern Oscillation (ENSO). By applying causal machine learning to remote sensing and socioeconomic data, they found that in regions where ENSO consistently reduces vegetation, a food price spike leads to a significant 5.4 percentage point increase in the population at acute risk in the subsequent month.

Conversely, in areas where vegetation is unaffected or positively influenced by ENSO, the impact of price spikes on food insecurity is considerably smaller and statistically insignificant (around 2 percentage points). These findings underscore the critical role of climate context in understanding vulnerability to food insecurity.

The study suggests that combining these sensitivity regimes with existing price-spike triggers can enable more effective anticipatory actions. By using ENSO states to identify vulnerable regions months in advance, aid organizations can pre-position responses, thereby mitigating the most severe spikes in acute food insecurity across sub-Saharan Africa.