Fine-grained prediction of food insecurity using news streams

TL;DR

This paper introduces a deep learning approach that analyzes news articles to predict food crises up to three months in advance, outperforming existing models and aiding humanitarian efforts.

Contribution

It presents a novel method leveraging causally grounded, interpretable text features from news streams for early food insecurity prediction at high spatial and temporal resolution.

Findings

Predicts 32% more food crises than existing models.

Achieves up to three months early warning at district level.

Validates the approach across 15 fragile states.

Abstract

Anticipating the outbreak of a food crisis is crucial to efficiently allocate emergency relief and reduce human suffering. However, existing food insecurity early warning systems rely on risk measures that are often delayed, outdated, or incomplete. Here, we leverage recent advances in deep learning to extract high-frequency precursors to food crises from the text of a large corpus of news articles about fragile states published between 1980 and 2020. Our text features are causally grounded, interpretable, validated by existing data, and allow us to predict 32% more food crises than existing models up to three months ahead of time at the district level across 15 fragile states. These results could have profound implications on how humanitarian aid gets allocated and open new avenues for machine learning to improve decision making in data-scarce environments.