Leveraging environmental microbial indicators in wastewater for data-driven disease diagnostics

Abstract

Introduction: Wastewater-based surveillance (WBS) is an emerging tool for monitoring the spread of infectious diseases, such as SARS-CoV-2, in community settings. Environmental factors, including water quality parameters and seasonal variations, may influence the prevalence of viral particles in wastewater. This study aims to explore the relationships between these factors and the incidence of SARS-CoV-2 across 28 monitoring sites, spanning different seasons and water strata.

Methods: Samples were collected from 28 sites, accounting for seasonal and spatial (surface and intermediate water layers) variations. Key physicochemical parameters, heavy metals, and minerals were measured, and viral presence was detected using RT-qPCR. After data preprocessing, correlation analyses identified 19 relevant environmental parameters. Unsupervised learning algorithms, including K-means and K-medoid clustering, were employed to categorize the data into four distinct clusters, revealing patterns of viral positivity and environmental conditions.

Results: Cluster analysis indicated that seasonal variations and water quality characteristics significantly influenced SARS-CoV-2 positivity rates. The four clusters demonstrated distinct associations between environmental factors and viral prevalence, with certain clusters correlating with higher viral loads in specific seasons. The clustering patterns varied across sample sites, reflecting the diverse environmental conditions and their influence on viral detection.

Discussion: The findings underscore the critical role of environmental factors, such as water quality and seasonality, in shaping the dynamics of SARS-CoV-2 prevalence in wastewater. These insights provide a deeper understanding of the complex interplay between environmental contexts and disease spread. By utilizing WBS and advanced data analysis techniques, this study offers a robust framework for future research aimed at enhancing public health surveillance and interventions.

Keywords: SARS-CoV-2; environmental factors; machine learning (ML); public healh; wastewater-based surveillance (WBS).

FIGURE 1

Overview of the study methodology, detailing the process from sample collection, through various stages of data processing, to final outcome representation. The diagram highlights key steps, including environmental sampling, laboratory analysis, data preprocessing, and clustering analysis for interpreting the results.

FIGURE 2

Heatmap represented Spearman’s rank correlation of all the physicochemical properties considered as independent features in this study. The correlation plots are depicted for (A) N gene, (B) ORF1ab gene, (C) RNaseP gene and (D) for RT-qPCR results. The plot showing the presence of both + ve and -ve correlation with the selected features.

FIGURE 3

The comprehensive plot of K- means clustering with distinct four clusters and their centroids. The cluster we obtained with respect to the two principal component PC1 and PC2 derived using principal component analysis (PCA).

FIGURE 4

SHAP dependence plot of the features (in terms of PC1 and PC2) of the k-means clustering model.

FIGURE 5

Variation in minerals and water quality parameters with season for changes in clusters. (A) variation in mineral content of sites for cluster change from 0/1 to 2 (B) variation in mineral content of sites for cluster change from 0/1 to 3 (C) variation in water quality parameters of sites for cluster change from 0/1 to 2 and (D) variation in water quality parameters of sites for cluster change from 0/1 to 3.