ABSTRACT
Background: The outbreak of Coronavirus disease 2019 (COVID-19) emerged in China at the end of 2019 and has spread worldwide within few weeks reaching a pandemic status. China and Italy, followed by other countries, have implemented strict control measures including complete lockdown. The effects of such measures and the optimal strategies to relax them are major Public Health concerns. Method: We introduce a SEIR compartmental model describing the COVID-19 pandemic, taking into account the region-specific fraction of undetected cases, the effects of mobility restrictions fitting Google COVID-19 Community Mobility Report data, and the personal protective measures adopted as frequent hand washing or use of masks. Results: The model is experimentally validated with data of Italian regions, European countries and US, obtaining fitting parameters in good agreement with previous literature. The mean absolute percentage error analysis for forecasting accuracy judgement showed that our model was highly accurate in 12 regions (46%), good in 7 (27%) and reasonable in 7 (27%). The estimation of the undetected cases allows to evaluate different post-lockdown scenarios which is crucial for public-health decision makers. Conclusions: The proposed model, calibrated and designed on measured or literature-based parameters, was successfully applied in all scenarios analyzed and the estimation of the dynamic of the whole number of infectious individuals could help the planning of different strategies aimed at relaxing the lockdown measures.