Source-linked AI summary
Mitigating Disease Spread by Design in Refugee and IDP Camps
Giulia Zarpellon, Joseph Aylett-Bullock, Frank Krauss, Miguel Luengo-Oroz
TL;DR
Disease spread is an increasing challenge in refugee and IDP settlements, and settlement layout influences movement and interactions relevant to transmission. The paper develops a JUNE-based methodology to compare prototypical camp layouts and facility placements, finding proof-of-concept differences in larger virtual settlements and outlining directions for further validation.
Problem
Disease spread threatens refugee and IDP settlements, while the potential of settlement layout as a mitigation strategy requires systematic exploration.
Method
The paper uses the JUNE agent-based epidemic model to simulate virtual camps with different geographies and facility-positioning schemes.
Results
In larger 16-region settlements, boundary placement produces lower infection curves than middle placement, while 4-region layouts show few substantial differences.
Takeaways & Limitations
The framework provides initial recommendations and operational pointers for investigating camp layouts as potential disease-spread mitigation factors.
Takeaways & Limitations
The proof-of-concept requires validation across multiple random seeds, systematic parameter exploration, physical-environment data, and additional contextual knowledge.
Abstract
from arXiv · showhide
Disease spread represents an increasing challenge in refugee and internally displaced person (IDP) settlements. The movement and interaction of people within camps is influenced by their layout, which therefore has the potential to significantly affect disease spread. This work aims at creating a methodology to explore the potential effects of different camp layouts as mitigating factors in the spread of diseases within settlements. We showcase proof-of-concept experiments by leveraging the JUNE agent-based epidemic model, discuss the kind of operational insights this methodology can facilitate, and provide a framework for future investigations.
1 INTRODUCTION
Disease spread is a growing threat in refugee and IDP settlements, where overcrowding and inadequate healthcare can accelerate outbreaks. Settlement geography and facility placement shape movement and interactions, creating an opportunity to mitigate spread through design.
- Overcrowding and insufficient healthcare facilities can accelerate disease spread in refugee and IDP settlements.
- Settlement geography and organization influence how people move and interact with one another.
- Placing shelters, homes, supermarkets, religious centers, schools, and other structures is therefore relevant to infection spread.
- Design paradigms that minimize disease spread could embed anticipatory mitigation in new settlements.
2 BACKGROUND
Camp planning combines conceptual design frameworks with spatial, resource, funding, and minimum-standard constraints. JUNE provides an agent-based modeling basis adapted to refugee settlements and calibrated to local mixing patterns.
- Planning refugee settlements: Camp design frameworks are refined in practice by spatial and geographical factors, resource and funding constraints, and minimum standards.
- Epidemic modeling with JUNE: JUNE originally modeled COVID-19 interventions in the UK using a high-granularity synthetic population and individual movements.
- Planning refugee settlements: The studied camps include grid-like structures documented in Zaatari, Qushtapa, and Wau.
- Epidemic modeling with JUNE: The refugee-settlement adaptation of JUNE was developed with UN agencies and calibrated for Cox’s Bazar using local and inter-camp mixing research.
3 OBJECTIVES AND METHODS
The paper develops a methodology that uses JUNE to compare how prototypical camp layouts and facility-positioning schemes may affect infectious-disease spread. It defines virtual geographies, synthesizes populations, distributes facilities, and simulates infection without containment strategies.
- The methodology explores camp layout as a potential mitigating factor and showcases proof-of-concept insights using JUNE in virtual camps.
- Defining a geography: Virtual settlements are defined with square camp building blocks subdivided into super-areas and areas, representing commonly observed grid-like modules.
- Generating a synthetic population: Synthetic populations are sampled from a selected basecamp’s demographic characteristics, with experiments presented using Zaatari and additional tests from Cox’s Bazar and Kismayo.
- Distributing shared facilities: Facilities are positioned using even, uniform, middle, or boundary schemes, while play groups and hand-pumps or latrines remain evenly distributed.
- Infecting the virtual settlement: The framework assesses layout effects without containment strategies such as social distancing.
4 EXPERIMENTS
Experiments compare virtual camps that differ only in venue positioning. Layout effects are small in four-region camps but become pronounced in larger settlements, where boundary placement produces lower infection peaks and fewer co-sharing users than middle placement.
- 58,860 people populate the 4-region setting and 241,957 people populate the 16-region setting in the Zaatari-based simulations.
- Infection curves: In 4-region camps, infection curves are comparable across venue-positioning schemes, whereas in 16-region camps boundary placement produces a substantially lower infection peak than middle placement.
- Agents’ mobility: In 16-region camps, evenly distributed venues yield smaller agent radius-of-gyration distributions than middle or boundary placement.
- Co-sharing users: Boundary placement produces fewer average co-sharing users, while middle placement increases the volume of people potentially sharing facilities.
5 DISCUSSION
The framework is a proof-of-concept methodology for assessing how facility layouts may affect disease spread in refugee and IDP settlements. Results suggest layout effects depend on settlement size, while further validation and richer contextual inputs are needed for operational use.
- Framework scope: The framework simulates epidemic spread in virtual camps with facilities distributed according to different prototypical positioning schemes.It is intended to provide initial recommendations and pointers for future investigations.
- Settlement size: In 4-region camps, infection spread rapidly saturates and dies out within camp boundaries, limiting differences between layouts.The discussion links this result to the limited spatial extension of these camps.
- Settlement size: In 16-region camps, boundary and middle facility schemes produce distinct infection patterns despite similar radius-of-gyration distributions.The boundary scheme is associated with lower infection curves than the middle scheme.
- Interpretation: The lower infection curves under the boundary scheme may be linked to agents sharing assigned venues with fewer people.This is presented as a possible mechanism rather than a confirmed causal explanation.
- Operational implications: Layout assessments may need repeated evaluation as camps evolve and expand, because venue positions can shift from boundary-like toward middle-like configurations.The paper frames this as part of continuous planning for camp evolution.
- Future work: The experiments require validation across multiple random seeds, systematic parameter exploration, and additional venue-positioning schemes across varied site shapes and topologies.Doubling per-capita rates, such as for learning centers, also affected infections and camp usage in initial experiments.
- Future work: Operational deployment would need to incorporate the physical environment and contextual constraints, including guidance that may discourage peripheral school locations.The framework is described as a first step toward a crisis-response tool for settlement design.
A DATA SOURCES AND MODEL ASSUMPTIONS
The model constructs virtual camp populations and venues from reference-camp demographics, facility rates, mobility assignments, and contact assumptions. It uses Zaatari-based population sampling, configurable facility placement, clustered infection seeding, and contact patterns informed by refugee-settlement data.
- Population data: The synthetic population samples age and sex from Zaatari camp distributions and uses reported household and inhabitant characteristics as source data.The source count was conducted by REACH and UNICEF between December 2014 and January 2015.
- Spatial assumptions: Virtual regions use fixed 150 m area sides and 0.81 km^2 per region rather than replicating the basecamp’s population density.Each region can accommodate up to 20,000 people while complying with the stated open-space guideline.
- Venue assumptions: Facility counts are determined from baseline per-capita parameters, with type-specific multipliers allowing scenarios containing different numbers of venues.Learning-center and play-group rates apply only to children ages 3–17.
- Infection seeding: The model seeds infection in 2 people per day for 10 days, concentrating initial cases in selected households and a particular region or super-area.This clustered scheme represents localized rather than uniformly distributed introduction of cases.
- Health assumptions: Co-morbidities are distributed using country-of-origin data and the approach described by Clark et al., selecting Syria for Zaatari.The passage identifies this as consistent with prior model applications.
- Interaction assumptions: Contact patterns are encoded in age- and setting-specific matrices informed by a Cox’s Bazar population survey and refined with JUNE.These matrices provide the interaction structure used by the epidemic model.
- Policy assumptions: The simulations do not test specific epidemic interventions or assume containment strategies, apart from hospitalization and sheltering for severely ill people.Severely ill individuals are assumed not to participate in other camp activities.
B RADIUS OF GYRATION
The radius of gyration characterizes each person’s spatial reach by weighting distances among assigned venues according to visit probabilities. It is smaller when utilized locations are close together and larger when they are farther apart.
- Radius of gyration measures an agent’s spatial reach within the camp from distances to assigned facilities.The measure uses the facilities assigned to each person by the model.
- For each person, the calculation uses assigned venues and distances from their positions to a probability-weighted center of mass.The venue-set definition and center-of-mass distance are specified for each person’s assigned venues.
- Visit probabilities weight each venue’s contribution and are derived from JUNE’s Poisson parameters, with probabilities equally distributed among venues of the same type.These probabilities depend on the person’s age and sex through the model’s predetermined behavior settings.
- Small radii indicate that a person uses nearby locations, whereas larger radii indicate that their utilized locations are more dispersed.
- The analysis includes all venue types except hospitals and assigns learning-center parameters of 1 for students and 4 for teachers.The values reflect one student shift and four teacher shifts per day at learning centers.
C ADDITIONAL EXPERIMENTS
Additional experiments extend the analysis across both settlement scales, examining infection curves, mobility distributions, sub-regional infection patterns, and co-sharing users. The 16-regions co-sharing analysis uses a 25% population sample for tractability.
- Figure 7 combines infection curves with radius-of-gyration distributions for the 4-regions settlement.
- Figure 8 disaggregates infection curves at sub-regional levels.
- Figure 9 shows the distribution of average co-sharing users for a 25% population sample in the 16-regions setting.The sample is used to make the computation tractable.