PERIL: a new platform for risk assessment and generation of insurance products
Pietro Carpanese, Luca Badin, Veronica Follador, Elisa Saler, Marco Donà, Francesca da Porto
Ultima modifica: 2025-07-21
Sommario
Italy is a highly susceptible country to catastrophic natural events, with the sixth highest costs incurred due to extreme events since 2000 according to Germanwatch (Global Climate Risk Index, 2021). Despite the fact that 75% of Italian buildings are exposed to at least one natural hazard, only 5% of them are currently insured against natural catastrophes (Gallagher Re, 2024). These numbers highlight the need to invest in risk awareness and in risk management processes carried out by insurance companies and policy makers. Therefore, this work aims to create a new platform that incorporates a multi-risk assessment framework, digitizing traditional risk analysis and tailoring models to the specific context of Italy.PERIL (Platform for the Evaluation of Risk: Insurtech innovation for Loss estimation) is a tool for estimating losses following natural disasters, particularly seismic and flooding events. This platform embeds up-to-date maps and models for the main natural hazards in Italy, vulnerability models for residential and industrial buildings, and it leverages computer vision algorithms to obtain detailed information on the buildings' characteristics from satellite and street view images. All these modules are combined within PERIL into a CatNat model for calculating risk in terms of losses for seismic and flood events. Lastly, an insurance module was added, which automatically returns possible insurance products, such as insurance policies and capital estimates needed to insure the entire building portfolio.Compared to software currently in use, PERIL makes it possible for insurance companies, public administrations, and risk management bodies to carry out risk analyses remotely and at different scales, allowing them to use all the models and algorithms for risk calculation in a simple, understandable, and automatic way.
è richiesta l'iscrizione al convegno per poter visualizzare gli interventi.