Estimating dynamic population served by wastewater treatment plants using location-based services data

Abstract

Wastewater-based epidemiology is a useful approach to estimate population-level exposure to a wide range of substances (e.g., drugs, chemicals, biological agents) by wastewater analysis. An important uncertainty in population normalized loads generated is related to the size and variability of the actual population served by wastewater treatment plants (WWTPs). Here, we built a population model using location-based services (LBS) data to estimate dynamic consumption of illicit drugs. First, the LBS data from Tencent Location Big Data and resident population were used to train a linear population model for estimating population (r2 = 0.92). Then, thespatiotemporal accuracyof thepopulation model was validated. In terms of temporal accuracy, we compared the model-based population with the time-aligned ammonia nitrogen (NH4-N) population within the WWTP of SEG, showing a mean squared error of < 10%. In terms of spatial accuracy, we estimated themodel-based population of 42 WWTPs in Dalian and compared it with theNH4-N and design population, indicating good consistency overall (5% less than NH4-N and 4% less than design). Furthermore, methamphetamine consumption andprevalence based on the model were calculated with an average of 111mg/day/1000 inhabitants and 0.24%, respectively, and dynamically displayed on a visualization system for real-time monitoring. Our study provided a dynamic and accurate population for estimating the population-level use of illicit drugs, much improving the temporal and spatial trend analysis of drug use. Furthermore, accurate information on drug use could be used to assess population health risks in a community.