Food Providers Deprived of Food: A Spatiotemporal Analysis and GeoAI Based Forecasting of Food Insecurity in Iran Rural Areas over the Recent Decade

Food insecurity in Iran has intensified in recent years due to high inflation, declining household purchasing power, and major reforms in food subsidy policies. Although food insecurity has long exhibited pronounced regional disparities, it has increasingly expanded into rural areas, yet conventional measurement approaches often fail to capture intra-national spatial heterogeneity. This study addresses this limitation by examining the spatiotemporal dynamics and spatial heterogeneity of food insecurity across Iran and forecasting its potential evolution using a GeoAI-based analytical framework. The analysis employs Foster–Greer–Thorbecke (FGT) indices derived from annual Household Income and Expenditure Survey (HIES) data for the period 2013–2024, while spatial clustering and local spatial dynamics are assessed using Moran's I and Local Indicators of Spatial Association (LISA). Results show that in 2013, approximately 39% of sampled households in 150 counties experienced food insecurity, largely concentrated in central, eastern, and southern regions, but between 2021 and 2023, the prevalence rose to over 60%, with its geographic extent expanding to 290 counties nationwide. The food insecurity gap—defined as the shortfall from the minimum daily requirement of 2,100 kilocalories—increased from 24% in 2013 to more than 27% in 2024, while the severity of food insecurity rose by approximately 13%. These trends coincided with subsidy policy revisions and changes in the preferential exchange rate for food imports, which triggered sharp food price increases. Projections to 2026 suggest that more than 60% of households may face food insecurity, highlighting the urgent need for economic stabilization measures and regionally targeted food security policies.
Introduction
Food insecurity, as a growing global challenge, reflects deep structural inequalities in access to healthy and affordable food. While much research has focused on urban areas, the rural dimension—critical for food production—remains underexplored. This multidimensional concept, encompassing availability, access, utilization, and stability, is difficult to measure. Despite the development of various indicators, no single global standard has yet been established. In Iran, although national food insecurity has somewhat decreased, over 3.5 million people still face severe food insecurity, with rural populations disproportionately affected due to lower incomes, limited market access, and greater dependence on agricultural livelihoods. Given recent socioeconomic shifts, such as the removal of food subsidies and rising inflation, there is a critical need for fine‑scale spatiotemporal analysis to capture local variations. Traditional survey‑based methods lack the capacity to capture local vulnerabilities, whereas spatiotemporal analytical frameworks and Geospatial Artificial Intelligence (GeoAI) offer powerful tools for modeling complex patterns and making accurate predictions. This study addresses this gap by calculating the FGT (Foster‑Greer‑Thorbecke) food insecurity index, identifying critical hotspots, analyzing spatiotemporal changes from 2013 to 2023, and predicting future patterns in rural districts of Iran using GeoAI.
Materials and Methods
This applied research employed a descriptive‑analytical approach utilizing Exploratory Spatial Data Analysis (ESDA). The primary data source was the raw Household Income and Expenditure Survey (HIES) data for rural households, collected annually by the Statistical Center of Iran from 2013 to 2023. Food consumption data were used to estimate daily per capita caloric intake, with the food insecurity threshold set at 2100 kilocalories per person per day. Food insecurity was measured using the FGT index (P₀, P₁, P₂), which captures prevalence, depth, and severity. Spatial distribution patterns and clusters were analyzed using Moran's I, Local Indicators of Spatial Association (LISA), and Hot Spot Analysis. Finally, a Long Short‑Term Memory (LSTM) neural network model was employed to predict food insecurity levels for the year 2026. The model's performance was validated using cross‑validation and the Root Mean Square Error (RMSE).
Results and Discussion
The findings reveal a clear and significant worsening of rural food insecurity in Iran from 2013 to 2023. The P₀ index (prevalence) increased from 39% to over 60% of rural households falling below the caloric threshold. Substantial spatial heterogeneity was observed, with prevalence in 2023 ranging from 21% in Semnan Province to nearly 88% in West Azerbaijan Province. Over the decade, Tehran Province had the highest average prevalence (~69%), while Kermanshah had the lowest (19%). The number of districts with severe food insecurity rose dramatically from 110 to 250. The P₁ index (depth) increased from 24% to 27%, indicating a decline in average caloric intake among food‑insecure households. The P₂ index (severity) grew by over 13% during the study period. Spatial analysis using Moran's I showed a decreasing concentration of food insecurity, indicating a shift from a clustered pattern to a more dispersed and widespread national issue. Early hotspots were concentrated in central and southeastern provinces but gradually expanded to northern and northwestern regions. In contrast, some western provinces consistently showed below‑average food insecurity levels. Predictions using the LSTM model for 2024–2026 suggest a continued high prevalence, with the P₀ index slightly exceeding 60%. However, scenario analysis indicates a wide range of possible outcomes: from a drastic increase to 82% in a pessimistic scenario to a reduction below 45% in an optimistic scenario. Projected spatial patterns for 2026 still show significant clusters of food insecurity in the northwest, north, center, south, and parts of the east, with the emergence of new hotspots.
Conclusion
The study confirms that caloric intake and access to sufficient food in many rural areas of Iran fall below standard levels, with significant spatial inequalities. While food insecurity showed a slight decline until 2017, policy shifts such as reduced government support in 2018 and the removal of food subsidies in 2022 led to a sharp increase in caloric deficits and insecurity, affecting over 250 districts. Income deprivation creates a detrimental cycle of reduced purchasing power, inadequate nutrition, and adverse social and health outcomes. Projections for 2026 indicate that persistent food insecurity may affect more than half of the country's districts. The findings underscore the urgent need for targeted, region‑specific interventions, price stabilization, and improved food subsidy policies.