Using a Model-Based Framework to Develop Wastewater Monitoring for Novel Infectious Disease Targets

Wastewater-based epidemiology (WBE) is an alternative, community-level approach to monitoring infectious diseases and informing public health responses. It bypasses the challenges of clinical surveillance methods, especially the need for individuals to get tested directly for accurate results. Yet, despite WBE’s demonstrated success with well-established targets such as poliovirus and SARS-CoV-2, applying WBE to emerging or neglected diseases remains largely unknown and challenging. The goal of my PhD research is to answer the question: Can WBE be feasibly applied to novel disease targets, and if so, what conditions are needed for successful WBE applications to occur? Part of the reason for the challenge in developing WBE for novel targets is that the current ad hoc or trial-and-error approach - conducting detection experiments without understanding whether a target is suitable for wastewater surveillance to begin with - can consume substantial resources and often yields results that are difficult to interpret without contextual knowledge of disease prevalence, shedding behavior, and analytical limitations. To address these challenges, I developed a mechanistic WBE feasibility framework that integrates biological, epidemiological, environmental, and analytical (i.e., disease, location, and laboratory-based) factors to assess the likelihood of successful wastewater detection for any given pathogen. By accounting for these conditions, this framework quantitatively evaluates the WBE feasibility of specific disease targets across a combination of conditions, demonstrating that there is no “one size fits all” approach for WBE when applied to different targets in different spatial, temporal, and methodological contexts. The framework provides several key applications. It can forecast WBE feasibility, predicting minimum infection rate requirements and optimal PCR replicate versus process limit of detection (PLOD) conditions before field deployment. It quantifies the importance of detection factors, including shedding rate, infection level, per capita wastewater flow rate, methodological detection limits, and the number of PCR replicates, and clarifies their influence on wastewater detection probability. It also identifies the dominant shedding pathways for each pathogen (feces, urine, saliva, etc.) and helps put past and future WBE successes and failures into perspective by contextualizing results according to methodological and location-specific factors. Importantly, the framework determines when, where, and how WBE should be implemented - the proper timing (infection levels), sampling location, and WBE methodologies (e.g., concentration methods and assay sensitivity) required. The framework can also be applied in conjunction with laboratory measurement results to assess the practicality of model predictions and quantify the relative effects of decay and super-shedding on wastewater signals, helping guide WBE applications more accurately. The model was first applied to emergent pathogens to demonstrate its predictive power. For mpox, it accurately forecasted that wastewater detection would be “feasible” (i.e., at 50% probability of detection) during the 2022 outbreak with a single PCR replicate if infection rates exceeded roughly 7 per 100,000 and if PLOD values and wastewater flow rates were relatively low. The prediction of mpox’s suitability for WBE was later confirmed by multiple independent studies successfully detecting mpox in wastewater worldwide. For Zika virus, the model predicted that detection would be challenging outside outbreaks or in non-endemic regions but potentially feasible in endemic countries such as Brazil and Colombia, where higher infection rates and lower per-capita wastewater flow rates enhance Zika wastewater detectability. Detection is inherently more difficult for Zika than mpox because its per-infection shedding is roughly 10-fold lower than mpox's, necessitating more fine-tuning of WBE protocols - namely, more sensitive workflows (i.e., lower PLODs) or a greater number of PCR replicates to achieve feasible detection. These predictions are consistent with the current literature, which reports detecting the virus in wastewater at low frequencies. Broadening the model to measles in a more global and historical context revealed that countries with very high case rates can achieve 50% detection using 2–3 PCR replicates at a sensitive PLOD of 3.0 log10 GC/L. In contrast, other regions require more replicates, more sensitive assays, or measles infection levels like those seen in pre-vaccination periods. Applying the framework to sexually transmitted infections (STIs) showed that HPV and chlamydia are promising WBE targets, while HBV, HCV, M. genitalium, and gonorrhea require specific conditions such as higher assay sensitivity or low per-capita flow; these quantitative predictions help guide prioritization and methodological optimization across this disease class. The malaria WBE study I conducted combines modeling with persistence and recovery spike-in microcosm experiments to evaluate the feasibility of detecting Plasmodium falciparum (the malaria parasitic agent) DNA in wastewater. The modeling predictions show that detection is most feasible in malaria-endemic Sub-Saharan African regions with a PLOD ideally below 3.0 log10 GC/L, and that DNA decay has minimal impact under typical wastewater conditions. Loss of DNA from freeze-thawing and low recovery, however, emphasize the need for optimized concentration and extraction methods. My pilot experiment using membrane filtration shows that the ideally low PLOD of 3.0 log10 GC/L can be achieved, demonstrating that the model predictions can be practically applied. The results of my PMMoV and Carjivirus comparative analyses show that predicted and observed wastewater loads (from the literature) and detection frequencies agree, confirming the robustness of the framework. Moreover, the analyses clarify the importance of super-shedding: failing to account for it can lead to overestimation of prevalence by 8.17- and 3.75-fold for PMMoV and Carjivirus, respectively, when the average shedding rate approach is used with WBE data. Both combined modeling–experimental analyses illustrate how mechanistic forecasting and empirical validation complement each other to enhance interpretability and predictive accuracy across WBE applications. The WBE feasibility framework is highly flexible and adaptable, supporting potential applications beyond what was demonstrated in this dissertation. For instance, future work could expand the framework to include wastewater surveillance of non-infectious health-related targets, such as chronic diseases, pharmaceuticals, and environmental contaminants. Furthermore, future studies collecting more diverse shedding data would enhance the model's prediction accuracy and comprehensibility. Moreover, integrating the framework with automated approaches, such as machine learning and online dashboards, could enable public health officials to assess, in real time, when, where, and how WBE can be applied to specific targets. Other potential future applications of the framework include assessing WBE feasibility in decentralized or building-level wastewater systems, accounting for temporal variability in wastewater flow, and integrating decay and recovery more comprehensively in a location-specific context by understanding how they mathematically relate to diverse environmental conditions. Ultimately, my work demonstrates that my WBE feasibility analysis framework can serve as a quantitative tool to guide broad wastewater surveillance across multiple targets and contexts. I hope my model can be expanded to enable future researchers to combat infectious diseases and other public health threats more effectively.<p></p>