A Spatially Validated Machine-Learning Framework for Mapping Alpha-Gal Syndrome Exposure Potential in Illinois
- 1University of Illinois Urbana Champaign
Abstract
The lone star tick, Amblyomma americanum, is closely associated with alpha-gal syndrome (AGS), an emerging tick-associated allergic condition. However, heterogeneous tick surveillance and limited clinical AGS data complicate the assessment of geographic exposure. We investigated whether ecological modeling and epidemiological surveillance could identify areas of potentially elevated AGS exposure across Illinois. Using data from 102 Illinois counties, eight statistical and machine-learning models were evaluated through geographic cross-validation to predict the relative abundance of lone star ticks. Random forest and XGBoost predictions were separately combined with reported ehrlichiosis rates to construct county-level AGS exposure-risk indices. Wildland-urban interface (WUI) population was examined in sensitivity analyses. Random forest achieved near-minimum prediction error while XGBoost demonstrated the strongest rank correlation. Both indices showed substantial concordance with CDC suspected-AGS categories across 97 counties (Spearman {rho} = 0.710 and 0.721). At the highest category threshold, odds ratios per 0.10-unit index increase were 5.02 and 3.63, respectively. By integrating ecological predictions with epidemiological surveillance, we establish a reproducible, spatially explicit framework for prioritizing AGS surveillance, targeting interventions, and guiding future epidemiological investigations across Illinois counties.
Related
- Machine Learning-Based Fatigue Detection from EEG Signals: A Methodological Pilot Framework with Potential for Nursing Applications
- WITHDRAWN: Multi-Timepoint Risk Stratification in Rare Cancers: A Computational Framework Validated against Published Ewing Sarcoma Trial Data
- Biopsy-Validated Deep Learning Mapping of Glioma Infiltration Beyond Contrast Enhancement Using Multiparametric MRI
- Needle-Free Bedside and Anthropometric Measures Show Weak Discrimination of an Adverse Blood-Biomarker Profile in Middle-Aged and Older Mexican Adults: An Internally Validated Machine-Learning Study
- Anatomy-guided U-Net deep learning for sparse-view chest CT reconstruction using spatially varying RTV regularization
BibTeX
@misc{hussain2026spatially,
title = {{A Spatially Validated Machine-Learning Framework for Mapping Alpha-Gal Syndrome Exposure Potential in Illinois}},
author = {A. Hussain and M.-P. Nohra and R. L. Smith},
howpublished = {medRxiv},
year = {2026},
month = jul,
doi = {10.64898/2026.07.24.26358812},
url = {https://www.medrxiv.org/content/10.64898/2026.07.24.26358812v2},
note = {CC BY},
}
RIS
TY - UNPB
TI - A Spatially Validated Machine-Learning Framework for Mapping Alpha-Gal Syndrome Exposure Potential in Illinois
AU - A. Hussain
AU - M.-P. Nohra
AU - R. L. Smith
PY - 2026
DA - 2026/07/24
PB - medRxiv
DO - 10.64898/2026.07.24.26358812
UR - https://www.medrxiv.org/content/10.64898/2026.07.24.26358812v2
AB - The lone star tick, Amblyomma americanum, is closely associated with alpha-gal syndrome (AGS), an emerging tick-associated allergic condition. However, heterogeneous tick surveillance and limited clinical AGS data complicate the assessment of geographic exposure. We investigated whether ecological modeling and epidemiological surveillance could identify areas of potentially elevated AGS exposure across Illinois. Using data from 102 Illinois counties, eight statistical and machine-learning models were evaluated through geographic cross-validation to predict the relative abundance of lone star ticks. Random forest and XGBoost predictions were separately combined with reported ehrlichiosis rates to construct county-level AGS exposure-risk indices. Wildland-urban interface (WUI) population was examined in sensitivity analyses. Random forest achieved near-minimum prediction error while XGBoost demonstrated the strongest rank correlation. Both indices showed substantial concordance with CDC suspected-AGS categories across 97 counties (Spearman {rho} = 0.710 and 0.721). At the highest category threshold, odds ratios per 0.10-unit index increase were 5.02 and 3.63, respectively. By integrating ecological predictions with epidemiological surveillance, we establish a reproducible, spatially explicit framework for prioritizing AGS surveillance, targeting interventions, and guiding future epidemiological investigations across Illinois counties.
ER -