Using maths to fight fraudulent motor insurance claims

Undetected fraudulent claims and lengthy processing times are in part the cause of rapidly deteriorating profitability in the global motor insurance market. It is estimated that fraudulent claims add approximately €50 to each premium in Ireland every year.

As of 2016, it was reported that roughly 1.95 million registered private cars were on the roads across Ireland and it can be assumed this number has continued to increase. However, it is estimated that 1 in 14 of these vehicles are driving without a valid insurance policy. Considering only the vehicles travelling Irish roads with valid insurance the annual loss to the motor insurance market from fraudulent claims is in excess of €90 million. Unfortunately until insurers can detect fraudulent claims with certainty, drivers' premiums will continue to cover this shortfall.

The decision making process currently used by insurance claims handlers relies heavily on reports from parties involved in the collision. The removal of this subjective reporting would alleviate a large proportion of fraudulent activity. Given the evolution of motor vehicles over the last decade, with almost all components now requiring a computer chip, is it possible to harness the data produced by such components to gain an impartial understanding of a vehicular impact?

Telematics is the technology used to send, receive and store data or information from remote objects. Vehicles with such technology fitted (all vehicles in the future will contain such technology) can sync data to the cloud allowing for information such as the geographical location and the acceleration of the vehicle throughout its journey be available to insurance handlers. Harnessing value from this data is where our research begins.

In conjunction with Irish based start up Xtract, the Mathematics Applications Consortium for Science and Industry (MACSI) at the University of Limerick have combined physics, mathematics and statistics to provide claims handlers with an unbiased visual reconstruction of vehicular impacts.

Commonly known for his discovery of gravity after watching an apple fall from a tree, Isaac Newton provided us with the fundamental mathematics we required - the second law of motion. The law states that the force exerted on an object (i.e. vehicle) is equal to the mass of the object multiplied by its acceleration, F = m ×a.

Using the known mass of the vehicle and the acceleration from telematics data, a calculation of the force exerted on the vehicle throughout its journey can be calculated. This allows us to identify the time of the impact using thresholding (i.e. a simple approach is to select the time point with the highest force), and the force of the impact.

Using the estimated time of impact, we were interested in detecting where on the vehicle the impact occurred. It is intuitive if the direction the vehicle travels directly after the impact can be estimated then the impact must have been directly opposite (e.g. if the car moves backwards after impact, then the car must have been hit from the front). Utilising the accelerometer data we developed an algorithm which identified which region of the vehicle had been impacted.

Armed with the impact time, the location of impact and the trajectory of the vehicle, the Xtract team built software which provides an animated reconstruction of the vehicle along its journey from pre-impact to post-impact for claims handlers to use when assessing claims.

The solution will not capture all €90 million of annual insurance fraud claims overnight, but we are sure that providing claims handlers with an unbiased reconstruction of impacts will aid in the decision making process. Hopefully in years to come as the number of years of your "no claims bonus" increases, your insurance premium will start to decrease/

This project won the Consultancy Impact award at the Knowledge Transfer Ireland (KTI) Impact Awards 2018 at a ceremony in Dublin Castle on 26th April.

Kevin Brosnan is an Irish Research Council-funded PhD researcher at MACSI in the Department of Mathematics and Statistics at the University of Limerick. Dr Sinéad Burke is manager of the MACSI at the University of Limerick

This article first appeard on RTE's Brainstorm website