Loading...
Overcoming barriers to telematics-driven risk assessment: AI approaches for driver behaviour analysis and claims prediction
Date
2025-09-30
Abstract
Transportation risk stakeholders, such as non-life insurers, can extract a vehicle's granular movements using a telematics device, thus predicting the driver's behaviours and risk potential. This data-driven approach enables the customisation of insurance policies, offering personalised premiums and promoting safer driving habits. Thus, the proliferation of telematics devices embedded in vehicles ensures a consistent stream of high-fidelity data for transportation risk stakeholders, enabling continuous improvement in risk assessment and policy development. Despite the availability of telematics datasets, research indicates that regulatory burdens and technological challenges hinder the effective use of these datasets. These obstacles create gaps in driver behavioural analytics, necessitating this research to investigate and address these issues. This thesis begins with a comprehensive review of telematics devices, focusing on their technical limitations, and the regulations governing ethical usage. Due to these limitations and requirements, transportation risk stakeholders face innovation inertia, particularly in adopting Artificial Intelligence (AI) models, as the black-box interpretability of these models fails to meet regulatory requirements and standards. In response to these challenges, this research promotes using inherently interpretable AI techniques and Explainable AI (XAI) methods. These approaches enhance model transparency, providing greater insight into vehicle risk and ensuring the ethical application of AI in insurance decision-making systems. Finally, the utility of combined telematics and claims data is showcased in a study which analyses the varying risk profiles of Electric Vehicles (EVs) and traditional Internal Combustion Engines (ICE) drivers. Overall, this thesis applies telematics as a conduit to investigate modern approaches to claim prediction, yielding greater insights into the rationale behind potential collisions and identifying improvements in analytical methods. Given these insights, transportation risk stakeholders such as insurers or national transport regulatory bodies have greater potential to encourage safer driver behaviours or reduce road collisions through fair pricing or improved road safety awareness campaigns.
Supervisor
Description
Publisher
University of Limerick
Citation
Collections
Files
Funding code
Funding Information
Sustainable Development Goals
External Link
Type
Thesis
Rights
http://creativecommons.org/licenses/by-nc-sa/4.0/