Fig 1.
Synthetic data generation workflow.
First, collected data from different sources is mapped in the domain geoJSON, generating gridded air quality data—after applying a dispersion model—and demographic data. This is combined with the dose-response functions from WHO to generate sequence activity data related with personal exposure and mortality relative risk.
Fig 2.
The first step of the IP is the preparation phase, where the simulations are performed; then, the generated data is processed through interactive process mining in constant discussion with domains expert, and the proposed scenarios are validated with domain experts once the KPIs are calculated again.
Fig 3.
Dashboard with base KPIs of the sample population.
In concrete, 24 h population exposure; mortality relative risk; percentage of risky activities and time spent in risky activities.
Fig 4.
ANOVA analysis for the exposure among activities.
The ANOVA table is shown in addition to the heatmap with the exposure difference between activities and the plot of the Tukey test.
Fig 5.
ANOVA analysis for the mortality relative risk among activities.
The ANOVA table is shown in addition to the heatmap with the mortality relative risk difference between activities and the plot of the Tukey test.
Fig 6.
Process map of the daily activities in the populations.
Redder nodes represent activities with a high spent time and green nodes show activities with a low spent time.
Fig 7.
Process map of the traces (citizens) with high exposures.
Highlighted nodes represent nodes with a significative difference in the spent time in comparison with the process map in Fig 6.
Fig 8.
Dashboard with KPIs for Scenario 1, in which population exposure and percentage of risky activities are reduced.
Fig 9.
Dashboard with KPIs for Scenario 2, in which population exposure, mortality relative risk and percentage of risky activities are reduced.
Fig 10.
Dashboard with KPIs for Scenario 3, in which population exposure is reduced.
Fig 11.
Dashboard with KPIs for Scenario 4, in which all KPIs are reduced.
Fig 12.
Dashboard with KPIs for Scenario 5, in which all KPIs are reduced in a greater percentage in comparison with Fig 11.
Table 1.
KPIs for different iterations of the model.
Fig 13.
IoT architecture.