Table 1.
Risk factors selected for simple and multiple logistic models for the prediction of dengue local transmission introduction probability. Model M1 (bold) was selected based on the result of AIC and residual deviance
| Model | Variable | Time lag (weeks) | AIC | Residual deviance | Odds | 95% CI | P-value |
|---|---|---|---|---|---|---|---|
| Null | 1 | 0 | 261.83 | 259.83 | – | – | |
| Simple | Dengue imported | 3 | 177.88 | 173.88 | 2.98 | 2.21–4.19 | <0.001 |
| Simple | Mosquito IMFA | 3 | 190.94 | 186.94 | 22.82 | 10.01–57.86 | <0.001 |
| Simple | Tweets | 3 | 191.32 | 187.32 | 1.04 | 1.03–1.05 | <0.001 |
| Simple | Minimum temperature | 4 | 231.22 | 227.22 | 1.26 | 1.15–1.38 | <0.001 |
| Simple | DENV in mosquitoes | 3 | 252.09 | 248.09 | 1.82 | 1.25–2.89 | <0.01 |
| Simple | Maximum temperature | 0 | 260.18 | 256.18 | 1.06 | 1.00–1.13 | |
| Simple | Mean temperature | 0 | 260.15 | 256.15 | 1.07 | 1.00–1.15 | |
| Simple | Humidity | 0 | 263.82 | 259.82 | 1.00 | 0.95–1.05 | |
| Simple | Rain | 0 | 263.82 | 259.82 | 1.00 | 0.94–1.06 | |
| M1 multiple | Dengue imported (lag 3) | 3 | 149.33 | 141.33 | 1.64 | 1.17–2.42 | <0.01 |
| Tweets (lag 3) | 1.02 | 1.01–1.03 | <0.01 | ||||
| Mosquito IMFA (lag 3) | 7.61 | 2.87–21.74 | <0.001 | ||||
| Multiple | Dengue imported (lag 3) | 3 | 151.00 | 141.00 | 1.62 | 1.16–2.39 | <0.01 |
| DENV in mosquitoes (lag 3) | 1.17 | 0.68–2.08 | |||||
| Tweets (lag 3) | 1.02 | 1.01–1.03 | <0.001 | ||||
| Mosquito IMFA (lag 3) | 7.51 | 2.83–21.48 | <0.001 | ||||
| Multiple | Dengue imported (lag 3) | 3 | 151.14 | 141.14 | 1.65 | 1.17–2.44 | <0.001 |
| Tweets (lag 3) | 1.02 | 1.01–1.04 | <0.01 | ||||
| Mosquito IMFA (lag 3) | 6.45 | 1.68–27.07 | <0.01 | ||||
| Minimum temperature (lag 4) | 1.03 | 0.88–1.20 | |||||
| Multiple | Dengue imported (lag 3) | 3 | 152.84 | 140.84 | 1.63 | 1.16–2.41 | <0.01 |
| DENV in mosquitoes (lag 3) | 1.17 | 0.68–2.07 | |||||
| Tweets (lag 3) | 1.02 | 1.01–1.03 | <0.01 | ||||
| Mosquito IMFA (lag 3) | 6.47 | 1.69–27.09 | <0.01 | ||||
| Minimum temperature (lag 4) | 1.02 | 0.88–1.20 | |||||
| Multiple | Tweets (lag 3) | 3 | 156.59 | 148.59 | 1.02 | 1.01–1.04 | <0.001 |
| Dengue imported (lag 3) | 1.98 | 1.41–2.87 | <0.001 | ||||
| Minimum temperature (lag 4) | 1.19 | 1.06–1.35 | <0.01 | ||||
| Multiple | Tweets (lag 3) | 3 | 156.93 | 150.93 | 1.03 | 1.02–1.04 | <0.001 |
| Mosquito IMFA (lag 3) | 13.46 | 5.50–36.37 | <0.001 |
Notes: Data indicate the influence of each variable studied in the response variable: 1 or more dengue confirmed local cases (local transmission of dengue introduction). Each variable was evaluated at 5 different time points of forecast (lag zero to four). The logistic model was performed for each unique variable. The odds column indicates how each variable affect the chance of dengue local transmission to occur. Orange marked variables were selected for further analyses
Abbreviations: AIC Akaike information criterion, CI confidence interval, IMFA mean infestation of female Aedes aegypti, DENV dengue virus