The information presented on this page is confidential and is made available solely for the use of reviewers and collaborators. Please do not share or distribute without permission.

---

On this page

• MAS
• West Nile Virus
• Chagas Disease

FloDisMod Multi-Agent System (MAS) Architecture

Based on: Zhao, Cerrone, Westerink & Dawson — Towards Multi-Agent Autonomous Reasoning in Hydrodynamics, arXiv:2605.01102 (2026)

Notional Layer Execution Graph (LEG)

💬  Natural-language query

→

🗺️  Graph Architect

→ constructs LEG from routing heuristics & specialist registry

Layer 1 — Hydrological specialists (parallel)

consolidator

Layer 2 — Ecological & disease specialists (parallel)

consolidator

reporter

↓

🗺️  Probabilistic risk maps

💬  Conversational insights

🚨  Early warning alerts

📂  Open datasets

Specialist Agent Registry

Hydrological Data

🌀

NHC

storm tracks
advisories

🌊

STOFS

surge forecasts
flood extent

📡

NOAA CO-OPS

water levels
tide gauges

📏

USGS

high-water marks
streamflow

🗂️

FEMA

flood zones
NFHL

🛰️

OSM / Image

basemaps
visual context

Ecological / Environmental

🌡️

Climate

temp · humidity
SST · rainfall

🌿

Land Cover

vegetation
urban extent

🏔️

Terrain / Bathy

DEM · LiDAR
standing water

🦟

Vector Habitat

breeding sites
niche detection

🌊

Sea Surface

SST · salinity
coastal plumes

Disease Specialists

🦟

Dengue

Aedes habitat
PDE mosquito sim
surveillance

🦠

Vibrio

SST · salinity risk
coastal water quality
STOFS-linked

🐛

Screwworm

Cochliomyia habitat
livestock exposure
agricultural data

Figure. Top: a notional Layer Execution Graph (LEG) for a representative query. The Graph Architect receives a natural-language query, consults the specialist registry, and constructs a query-specific LEG. Colored boxes in the LEG represent unnamed specialist agents drawn from the registry; colors correspond to agent type. A consolidator fuses parallel outputs between layers; the reporter synthesizes the final response with full provenance logging. Bottom: the full specialist registry from which the Graph Architect selects agents for any given LEG. Hydrological agents (blue) connect to NHC, STOFS, NOAA CO-OPS, USGS, FEMA, and OSM for flood and storm data. Ecological agents (green) supply climate forcing, terrain, land cover, vector habitat, and sea-surface conditions that define transient ecological niches. Disease specialist agents handle Dengue (Aedes habitat model; PDE-based mosquito simulation validated against Hurricane Harvey trap counts), Vibrio (sea-surface temperature and salinity risk linked to STOFS operational output), and Screwworm myiasis (Cochliomyia hominivorax habitat and agricultural exposure data).

---

Forecasting of Climate-Sensitive Mosquito Distributions to Evaluate Vector-Borne Disease Risk

The increased occurrence of extreme weather events is reshaping the distribution of disease vectors, increasing the risk of emerging diseases and expanding the transmission zones for pathogens such as West Nile virus (WNV). Initially, focusing on North America, we are developing a scalable framework to model and forecast the distribution of key mosquito vectors, known to be responsible for the transmission of WNV. By integrating hydrological and ecological niche modeling, landscape description and climate projections, we generate interpretable predictions of current and future habitat suitability of mosquito species. To complement this framework, initially, we explicitly focus on *Culex pipiens pipiens* and *Culex pipiens quinquefasciatus*, two closely related, often hybridizing mosquito species. They are major vectors for WNV in North America: *C. p. pipiens* is generally found in temperate, higher-latitude regions and *C. p. quinquefasciatus* is dominant in tropical and subtropical regions. Figure 1 shows the occurrence map of *C. p. pipiens* and *C. p. quinquefasciatus* mosquito species in North America between 2020–2025. In total, 4981 occurrence records were identified.

**Figure 1:** Occurrence map of *C. p. pipiens* and *C. p. quinquefasciatus* mosquito species in North America between 2020–2025 (N=4981). We separately modeled the habitat suitability of each mosquito species independently using MaxEnt in R and subsequently projected the habitat suitability under multiple climate change scenarios (SSP1, SSP2, SSP3, and SSP5) for the period 2070–2100. The resulting maps (Figure 2) illustrate the projected suitability of *C. p. pipiens* (CPP) and *C. p. quinquefasciatus* (CPQ) mosquitos under SSP1, SSP2, SSP3 and SSP5 scenarios, for the period 2070–2100.

**Figure 2:** Projected suitability of *C. p. pipiens* (CPP) and *C. p. quinquefasciatus* (CPQ) mosquitos under SSP1, SSP2, SSP3 and SSP5 scenarios, for the period 2070–2100. By integrating these projections with the 2022 county-level Social Vulnerability Index (SVI) across USA (Figure 3), this approach provides foundation and support for climate-informed public health planning through early warning systems, targeted vector surveillance, and adaptive control strategies.

**Figure 3:** Overlay of *C. p. quinquefasciatus* projection of its habitat suitability under climate change scenario SSP5 (2070–2100) with the Social Vulnerability Index for USA in 2022.

---

Tracking the Hidden Spread: Predicting Chagas Disease Risk Through Triatomine Habitat Models in North America

Triatomines (kissing bugs) are the primary vectors responsible for spreading the parasite *Trypanosoma cruzi* that cause Chagas disease. We present a new curated dataset of triatomine observations in the Americas containing 24,933 observations sourced from literature, citizen science, and new unpublished data. The entire dataset contains 127 species, of which 35 species are from North America and represent 62.0% of the data reported here. After cleaning and filtering, data from 14 species in North America were identified to have sufficient observations for calculating habitat suitability maps (HSMs) and projecting species distributions under different climate change scenarios. Our HSMs support the endemicity of Chagas Disease in the USA and the potential for spread into Canada towards the end of this century. This curated collection of triatomines and HSMs can assist in understanding potential exposure risks and guide ongoing screening strategies for Chagas disease in North America.

**Figure 4:** Workflow for Development of Habitat Suitability Maps of Each Kissing Bug (Triatomine) Species in Current and Future Climate Scenarios.

**Figure 5:** Geospatial Habitat Distribution of 16 Predominant Kissing Bug Species in North America. --- Back to Research