Predictive Water Virology: Hierarchical Bayesian Modeling for Estimating Virus Inactivation Curve

Abstract

Hazard analysis and critical control point (HACCP) are a series of actions to be taken to ensure product consumption safety. In food poisoning risk management, researchers in the field of predictive microbiology calculate the values that provide minimum stress (e.g., temperature and contact time in heating) for sufficient microbe inactivation based on mathematical models. HACCP has also been employed for health risk management in sanitation safety planning (SSP), but the application of predictive microbiology to water-related pathogens is difficult because the variety of pathogen types and the complex composition of the wastewater matrix does not allow us to make a simple mathematical model to predict inactivation efficiency. In this study, we performed a systematic review and meta-analysis to construct predictive inactivation curves using free chlorine for enteric viruses based on a hierarchical Bayesian model using parameters such as water quality. Our model considered uncertainty among virus disinfection tests and difference in genotype-dependent sensitivity of a virus to disinfectant. The proposed model makes it possible to identify critical disinfection stress capable of reducing virus concentration that is below the tolerable concentration to ensure human health.

Regularized regression analysis for the prediction of virus inactivation efficiency by chloramine disinfection

“…The prediction results of the coxsackievirus and echovirus models implied that the strain type was an important factor for predicting enterovirus LRVs (Fig. 4), so the hierarchical Bayesian approach, 21 which took more than three types of strain into account, was likely to improve the LRV prediction for the poliovirus. In addition, it is possible to take into account the effect of experimental conditions, which are not recorded in the articles on the LRV prediction by the approach.…”

“…In addition, it is possible to take into account the effect of experimental conditions, which are not recorded in the articles on the LRV prediction by the approach. 21 The probability distributions of the LRV of each virus species were determined based on AIC (Table S2 †). The variables selected by the best fit models based on the regularized regression analyses (Table 4) were applied here.…”

“…In this study, we have focused on chloramine disinfection and proposed predictive inactivation models for enteric viruses using the predictive water virology approach with water quality information as explanatory variables, which makes it possible to determine the CL to achieve the required disinfection efficiency under site-specific water quality. 21 Chloramine often exists in treated wastewater since free chlorine is converted to chloramine by reacting with ammonia. Chloramine has a virucidal activity and persists in water longer than other disinfectants, such as free chlorine and ozone.…”

Predictive water virology using regularized regression analyses for projecting virus inactivation efficiency in ozone disinfection

“…Thus, the disinfection intensity that achieves the target LRV needs to be determined by taking water quality and operational information at each WWTP into account. Previously, we proposed the concept of predictive water virology, in which the models predicted virus LRV by using water quality and operational parameters as explanatory variables ( Kadoya et al., 2019 ). The predictive water virology enables us to derive the appropriate CL to attain the target LRV under site-specific water quality.…”

“…The predictive water virology enables us to derive the appropriate CL to attain the target LRV under site-specific water quality. Kadoya et al. (2019 and 2020) previously proposed the concept of predictive water virology and reported that the predictive inactivation models, based on hierarchical Bayesian modeling (HBM) and regularized regressions, flexibly correspond to the changes and variety in water quality parameters among WWTPs ( e.g., seasonal variation).…”