Year: 2026
Type: article
Abstract: Accurately detecting infectious disease surges is essential for timely public health response. We asked whether incorporating SARS-CoV-2 viral lineage information improves machine learning–based detection of COVID-19 surges. We obtained COVID-19 clinical case counts and wastewater collected from 17 wastewater treatment plants’ catchment areas and viral lineage data from March 2022 to February 2023 in Minnesota. We developed a two-step framework in which surge periods were first defined, followed by training a supervised classification model to identify surge weeks. Model performance was evaluated using nested leave-one-group-out cross-validation (LOGOCV), holding out wastewater treatment plants to assess generalization to unseen locations within the study period. A model using wastewater viral load measures alone achieved a recall of 0.57, precision of 0.27, F1-score of 0.33, area under the precision-recall curve (PR-AUC) of 0.38, and accuracy of 0.65. Adding a lineage-derived feature increased recall to 0.72 (0.15 improvement, 95% CI 0.05 – 0.25), precision to 0.41 (0.14 improvement, 95% CI 0.07 – 0.22), F1-score to 0.48 (0.15 improvement, 95% CI 0.07 – 0.23), PR-AUC to 0.71 (0.33 improvement, 95% CI 0.22 – 0.44), and accuracy to 0.76 (0.11 improvement, 95% CI 0.06 – 0.17). These findings demonstrate that integrating genomic surveillance with wastewater data can meaningfully enhance infectious disease surge detection. The proposed framework may be generalizable and could potentially be applied to future pathogens.
Institutions: University of Minnesota, Minnesota Department of Health
Language: English
Subfield: Infectious Diseases
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,625
Awards: G2566360032, G4187292413
OpenAlex ID: W7234742897
