Fig 1.
Timeline of observed Ae. aegypti breeding sites in Geylang, Singapore, August 2014 –August 2015.
There were 107 days of entomological observations where 53 breeding sites, once identified as positive for Ae. aegypti breeding, were monitored for changes in hydrological conditions. Breeding sites where water exceeded the drainage threshold were classified as “Flushed”. For each day of observation, if at least one breeding site was observed as flushed, the day of observation was defined as flushed. If no breeding sites were observed as flushed then the day of observation was classified as not flushed. This figure, “Timeline of observed Ae. aegypti breeding sites in Geylang, Singapore, August 2014 –August 2015 is a derivative of “Timeline of the breeding drains of Aedes aegypti in Geylang, Singapore: August 2014 –August 2015” by Seidahmed and Eltahir [34], used under CC BY.
Table 1.
Rainfall variables created for PLUM model development.
These variables were created to characterize rainfall several different ways in order to capture the different mechanisms by which storm drains may be flushed.
Table 2.
Performance of PLUM model classification on the unseen test data.
The model used the aggregate number of high risk thresholds that were met per day for both cumulative and daily rainfall variables. Evaluation measures include F1-score (F1), accuracy (Acc), positive predictive value (PPV), Sensitivity (Se), area under the receiver operating characteristics curve (AUC), specificity (Sp), negative predictive value (NPV).
Fig 2.
Results of PLUM model associating flushing with the number of high risk thresholds met for cumulative and daily rainfall variables.
There is a clearly defined threshold that almost perfectly separates flushed (blue) and non-flushed (orange) observations based upon the aggregate number of high risk thresholds that were met per day for both cumulative and daily rainfall variables. Each gray line represents Eq (1) fit to each leave-one-out cross validation sample while the blue line represents the mean fit from all of the leave-one-out cross validation samples.
Table 3.
Descriptive statistics for weekly data on weather variables and dengue outbreak occurrence in Singapore, 2000–2016.
Fig 3.
The average prevalence of flushing events and dengue outbreak weeks by month, Singapore, 2000–2016.
In months where flushing event prevalence is highest (blue) dengue outbreak week prevalence (red) is lowest, and vice versa.
Fig 4.
Association between number of flushing events per week and dengue outbreak occurrence over 20 lag weeks.
The risk of an outbreak occurring within one to six weeks after the week of consideration was significantly lower if five or more flushing events occurred during the considered week.
Table 4.
Association between number of flushing events per week and dengue outbreak occurrence over 20 lag weeks.
When five or more flushing events occurred in a week, compared with 0 flushing events, the risk of a dengue outbreak occurring in the subsequent weeks was significantly reduced up to six weeks after the flushing events occurred.