[TMP-087] Multivariate Asynchronous Shapelets for Imbalanced Car Crash Predictions
Car crash detection enhancement using explainable AI and CDR data, improving model transparency and decision-making
The growing availability of real-time sequential data and AI decision-making systems is transforming the mobility industry. Crash Data Recorders (CDRs), typically installed on airbag control modules, collect data before and after crashes to monitor safety and vehicle status. Leveraging machine learning (ML), these devices are now valuable for research and business applications, including customer service improvements for insurance companies.
In collaboration with Generali Italia, Italy’s largest insurance provider, this work supports an AI-driven system that uses CDR data to detect potential car crashes and alerts operators enabling informed decisions on whether to contact customers. Generali offers CDRs to customers, collecting data like speed and acceleration to train the system's deep learning model.
However, two challenges exist: high system sensitivity causing unnecessary calls and the opaque nature of deep learning models ("black-box"), which hinders trust and model refinement. Explainable AI (XAI) is crucial to address these issues by interpreting model predictions and improving decision-making.
This project uses real-world time series data for two tasks: standard classification and highly imbalanced classification, akin to anomaly detection. We combine post-hoc and ante-hoc XAI methods for transparency and introduce Multivariate Asynchronous Shapelets, an interpretable-by-design approach to enhance Generali’s predictive system and outperform state-of-the-art classifiers.
A pipeline combining post-hoc and ante-hoc XAI for standard time series classification and the introduction of Multivariate Asynchronous Shapelets, an interpretable method developed to surpass state-of-the-art classifiers and Generali’s black-box model. The results are published.
In addition to the scientific contribution on XAI, it is important to highlight how the application of AI systems to automate the remote detection of potential car accidents by an insurance company has a positive impact on road safety, improving rescue operations and helping to reduce the potential impacts of an accident on the health of the insured.
Partners
- Università di Pisa
- ISTI-CNR Pisa
- Generali Italia