Estimating sewage flow rate in Jefferson County, Kentucky, using machine learning for wastewater-based epidemiology applications

Abstract Direct measurement of the flow rate in sanitary sewer lines is not always feasible and an important parameter for normalization data used wastewater-based epidemiology applications. Machine learning to estimate past wastewater influent rates supporting public health applications has been studied. The aim this study was assess treatment plant when compared with weather retrospectively Louisville, Kentucky (USA), based on other data-types using machine learning. A random forest model trained a range variables, such as feces-related indicators, that could be associated dilution sewage systems, area demographics. developed algorithm successfully estimated accuracy 91.7%, although it did perform well short-term (one-day) high rates. This suggests variables precipitation (mm/day) population size are more estimation. fecal indicator concentration (cross-assembly phage pepper mild mottle virus) less important. Our challenges currently accepted opinions by showing potential application artificial intelligence estimation epidemiological