Abstract
Since a resurgence occurred in 1993, malaria has remained an endemic disease in the Republic of Korea (ROK). A major challenge is the inaccessibility of current vector mosquito abundance data due to a 2-week reporting delay, which limits timely implementation of control measures. We aimed to nowcast mosquito abundance and assess its utility by evaluating the predictive value of mosquito abundance for malaria epidemic peaks. We used machine learning models to nowcast mosquito abundance, employing gradient boosting models (GBMs), extreme gradient boosting (XGB), and an ensemble model combining both. Various meteorological factors served as predictors. The models were trained with data from mosquito collection sites between 2009 and 2021 and tested with data from 2022. To evaluate the utility of nowcasting, we calculated the effective reproduction number (R (t)), which can indicate malaria epidemic peaks. Generalized linear models (GLMs) were then used to assess the impact of vector mosquito abundance on R (t). The ensemble models demonstrated the best performance in nowcasting mosquito abundance, with a root mean square error (RMSE) of 0.90 and R-squared value (R (2)) value of 0.85. The GBM model showed an RMSE of 0.91 and R (2) of 0.84, while the XGB model had an RMSE of 0.92 and R (2) of 0.85. Additionally, the R (2) of the GLMs predicting R (t) using mosquito abundance 2 weeks in advance was >0.72 for all provinces. The mosquito abundance coefficients were also significant. We constructed reliable models to nowcast mosquito abundance. These outcomes could potentially be incorporated into a malaria early warning system. Our study provides evidence to support the development of malaria management strategies in regions where malaria remains a public health challenge.