As the world grappled with the unprecedented challenges brought on by the COVID-19 pandemic, one glaring issue emerged: the inability of aid organizations to swiftly and accurately identify the most vulnerable households in need of assistance. Traditional methods, reliant heavily on sociodemographic data, often failed to capture the rapidly shifting realities, leaving many deserving families stranded without essential support. This predicament underscored a fundamental flaw in established poverty assessment frameworks, particularly in urban contexts where deprivation manifests unevenly and dynamically.
In response to these challenges, Woojin Jung, an assistant professor at the Rutgers School of Social Work, embarked on pioneering research to revolutionize poverty targeting strategies. Collaborating with academic partners from the University of Bordeaux in France and Istanbul Technical University in Turkey, Jung’s team crafted an innovative hybrid methodology integrating household surveys, community perceptions, and state-of-the-art geospatial technology. This approach was applied in Lusaka, Zambia’s rapidly expanding capital with a population of nearly 4 million, serving as a vital testbed for refining aid distribution methodologies in complex urban environments.
Traditional poverty measurement tools have predominantly depended on quantifiable household characteristics such as asset ownership and income brackets. While foundational, these metrics alone often fail to account for the nuanced, localized experiences of marginalization and food insecurity within urban neighborhoods. Jung’s research challenges this paradigm by incorporating community members’ insights—open-ended feedback on neighborhood wealth and deprivation—placing affected populations at the heart of data gathering. These subjective expressions reveal underlying factors such as access to public utilities, personal living standards, and neighborhood infrastructure that raw numerical data may overlook.
Central to this methodological advancement was the integration of publicly available geo-spatial datasets, including high-resolution satellite imagery and luminosity maps. These tools enable researchers to quantify environmental and infrastructural variables with unprecedented granularity, linking observable spatial features to socioeconomic conditions. By applying artificial intelligence and machine learning algorithms, the research team could detect patterns and correlations between satellite-derived indicators and household-level poverty markers, carving a path toward more precise, scalable poverty predictions.
What sets Jung’s research apart is the ambition not merely to predict conventional wealth scores but to extend beyond and capture complex measures of food consumption and nutritional outcomes. These dimensions are notoriously challenging to quantify yet critical in understanding multifaceted poverty. Urban poverty is heterogeneous, with adjacent households often experiencing starkly different levels of deprivation due to inconsistent access to utilities, services, and infrastructure. Jung’s models aspire to predict these fine-grained disparities at the household level, advancing beyond the “small area estimates” that aggregate data across wider zones, thereby enhancing the resolution and effectiveness of aid targeting.
The data collection phase involved surveying 300 households in Lusaka, gathering both traditional wealth indicators—such as appliance ownership and frequency of iron-rich food consumption—and rich qualitative data from community members. The respondents’ perceptions illuminated aspects of neighborhood viability often invisible to external observers, including the reliability of health services, electricity availability, and the broader urban fabric shaping residents’ daily realities. These insights, when fused with satellite-derived spatial variables, formed a powerful composite dataset.
Applying sophisticated machine learning models on this integrated data, the research discovered that hybrid approaches combining household-level demographics with community-informed spatial indicators consistently outperformed standard methods that relied solely on conventional data. Remarkably, this blend reduced errors in simulated targeting of aid by as much as 26 percent, a significant margin that could translate into millions of dollars saved and thousands of households effectively reached in real-world scenarios.
An intriguing and transformative finding was the capacity to accurately identify multiply deprived households using only geospatial features. This highlights the potential of remote sensing to serve as a low-cost, rapid assessment tool in settings where traditional survey data is sparse or outdated. If generalized to other urban contexts, this methodology could revolutionize how governments and aid organizations deploy resources, moving toward a more data-driven, equitable framework.
The implications of this research extend beyond technical innovation. It underscores a vital paradigm shift towards participatory development, where beneficiaries’ voices and experiences actively shape intervention strategies. In an era where artificial intelligence increasingly informs policy and humanitarian actions, grounding technological tools in community engagement is essential to build trust, legitimacy, and effectiveness.
This collaborative effort with the Zambian government illustrates the real-world applicability of such hybrid models. The government plans to leverage these insights to refine agricultural input assistance—such as the targeted distribution of seeds and fertilizers—to support vulnerable farming households more efficiently. This partnership highlights the promise of combining academic rigor with practical governance to build stronger social safety nets, particularly as global foreign aid diminishes.
Dr. Jung emphasizes that impactful policy change transcends methodological tweaks; it necessitates earning the confidence of the people served. Making communities integral to the solution fosters a sense of ownership and ensures that aid delivery resonates with actual needs on the ground. In AL urban environments marked by complex socioeconomic tapestries, this inclusive approach may prove decisive in breaking cycles of poverty and food insecurity.
As urbanization accelerates worldwide, the need for precise, responsive poverty mapping grows ever more urgent. Jung’s research offers a beacon of hope, showcasing how cutting-edge technology, combined with human-centered insights, can transform humanitarian aid from a blunt instrument into a finely calibrated mechanism of social justice and resilience.
The publication of these findings in the esteemed journal Sustainable Cities and Society signals growing recognition of the convergence between social science, geospatial analytics, and machine learning in tackling pressing global challenges. Moving forward, scaling such methodologies and fostering similar partnerships across diverse cities will be imperative to achieving inclusive urban development and ensuring no vulnerable household remains invisible or unsupported.
Subject of Research: People
Article Title: Targeting urban poverty and food insecurity: A community-informed spatial analysis and machine learning approach
News Publication Date: 12-Sep-2025
Web References:
- Rutgers School of Social Work
- Responding to COVID-19 Pandemic – UNDP Report
- Benefits and Risks of AI for NGOs
- ScienceDirect Journal Article
References: DOI: 10.1016/j.scs.2025.106799 (Sustainable Cities and Society)
Keywords: Poverty, Urban Poverty, Food Insecurity, Machine Learning, Satellite Imagery, Social Science, Aid Targeting, Community Participation, Geospatial Analysis
