Key points are not available for this paper at this time.
ABSTRACT Overview graphic showing the community-level data inputs into the random forest model used to predict school attendance. Community-transmitted seasonal infectious diseases pose a significant challenge to school systems across the nation. School attendance is central to student learning progress and is a key public health indicator. We aimed to evaluate the incremental explanatory value of community-level seasonal influenza wastewater concentrations for estimating school absenteeism within the corresponding geographic sewershed in Louisville, Kentucky (USA). We used community wastewater seasonal influenza concentrations from a catchment area that geographically encompassed 51 public elementary, middle, and high schools. These data were compared with daily school-level attendance records provided by Jefferson County Public Schools for three academic years (2022–2025). Two random forest models were developed that incorporated community influenza A wastewater concentrations, environmental variables (e.g., rainfall and maximum temperature), school-level attendance, and total enrollment to predict attendance outcomes. Across the sewershed, the all-schools model (N = 51) performed well (training R2 = 0.94, testing R2 = 0.60, and MAE = 2.88); however, the high school-only model (N = 7) performed better (training R2 = 0.98, testing R2 = 0.86, and MAE = 3.22). Our findings present a preliminary approach for using wastewater-based epidemiology and machine learning to enhance early-warning capacity and inform health-supportive responses across educational settings.
Dusterhoff et al. (Mon,) studied this question.