System and method for predicting accident damage estimates for cars

Abstract

The description relates to a system and method for predicting accident damage estimates for cars. More particularly, the description relates to a system and method for predicting accident damage estimates for new cars that utilizes accident data pertaining to previous insurance claims and car assembly modeling to predict accident damages. These accident damage predictions are used as input to calculate appropriate insurance premiums.

Description

SYSTEM AND METHOD FOR PREDICTING ACCIDENT DAMAGE ESTIMATESFOR CARSFIELD OF THE INVENTION

[0001] The description relates to a system and method for predicting accident damage estimates for new cars. More particularly, the description relates to a system and method for predicting accident damage estimates for new cars that utilizes accident data pertaining to all cars and car assembly modeling to predict accident damages.DESCRIPTION OF THE PRIOR ART

[0002] The make, model, year, value and potential cost to repair a car all impact the cost of insurance. The cost to insure a car is subject to change for many reasons. For example, some vehicles are more susceptible to theft, while other cars may be better designed to, include more safety features and are less likely to sustain serious damage due to a collision. Insurance rates are also affected by the cost to repair vehicles, which changes based on each make and model. Further, some cars fare better than others in collisions, resulting in fewer injuries and minimal car damage, which results in lower insurance rates. The cost of insurance also accounts for the car’s replacement value and the depreciation of the same over time. Insurance companies use previous claim data to rank automobiles according to risk and use this to determine appropriate premiums. This ranking of claims data is shared publicly in Canada in a table known as the CLEAR table. The CLEAR table shows you the relative cost of repairing cars, passenger vans, SUVs, trucks and wagons and this data is used when calculating insurance premiums. The CLEAR table is compiled based on a large number of claims per vehicle make and model. As a result, cars without a significant number of claims either do not appear on the CLEAR table or the data associated with a car with too few claims would be inaccurate. Thus, a new make and model of vehicle must have been available to the public and on the road for a significant amount of time before accurate information can be presented on the CLEAR table.

[0003] When new cars are released, they can not be included on the CLEAR table as there are not 1500 insurance claims. Therefore, companies are blind to the cost of repairs, and it is therefore difficult to price those policies because no accidents have occurred. Whilethe insurance companies can base their premiums on new cars based on the previous year of the same model, this is not always an accurate estimate of appropriate premiums for the new model. Each model year has different parts arranged in a different manner and the cost of repairs can be drastically different between model years. There remains a need for a method to accurately predict appropriate insurance premiums for new car models.SUMMARY OF THE INVENTION

[0004] The description relates to a system and method for predicting accident damage estimates for cars. More particularly, the description relates to a system and method for predicting accident damage estimates for new cars that utilizes accident data pertaining to previous insurance claims and car assembly modeling to predict accident damages. These accident damage predictions are used as input to calculate appropriate insurance premiums.

[0005] The description relates to a method for determining insurance premiums for a vehicle comprising the steps of a processor accessing a database containing accident claim data and calculating a predicted accident profile. The predicted accident profile contains different types of accidents. The accident profile is used as input into estimation modeling software to determine an approximate repair cost factor based on determined accident type. The processor uses the approximate repair cost factor to determine appropriate insurance premiums for the vehicle.

[0006] In a further embodiment, the vehicle is a new make or model of vehicle.

[0007] In yet a further embodiment, the accident profile further comprises weighting factors associated with each of the determined accident types. These weighting factors are associated with each of the accident types being used to determine the repair cost factor. The repair cost factor is determined by calculating a weighted average of the cost of repair of each accident type in the accident profile.

[0008] In yet a further embodiment, the estimation modeling software uses deformation estimates obtained from the database to determine the parts likely to require replacement, the cost of parts likely to require replacement, and the approximate cost of labour to replace parts likely to require replacement. This is used as input to calculate the approximate cost of repair for each accident type.BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The features of the invention will become more apparent in the following detailed description in which reference is made to the appended drawings wherein:

[0010] Figure 1 is screen shot from estimation modeling software of the prior art;

[0011] Figure 2 is a flow chart showing one embodiment the proposed system and method described herein;

[0012] Figure 3 is a flow chart showing a second embodiment the proposed system and method described herein; and

[0013] Figure 4 is a flow chart showing a third embodiment the proposed system and method described herein.DETAILED DESCRIPTION OF THE INVENTION

[0014] Insurance companies have, over the years, collected a large volume of data regarding automotive based insurance claims. From this data, information regarding collisions including but not limited to, type of accidents and deformation of the vehicle for each type of accident can be determined. Types of accidents can be determined based on many characteristics, for example, driving location, type of vehicle, and age of driver.

[0015] Currently, when a vehicle is involved in an accident, an appraiser takes photos of the vehicle and uses them to estimate the damage. Computer software currently exists where vehicle components are modeled and based on areas of the vehicle, appraisers can click on the parts that were affected in the accident. The software is linked to a database of part prices / availability allows appraisers to quickly come up with estimates for the repair of the vehicle. An example screen shot 100 from an estimation modeling software is shown in figure 1. In this example, components of the front hood require replacement. If front hood 102 requires replacement, an adjustor clicks on the front hood and indicates if a replacement part is required, if the part requires repair, or if another option to fix the damage is required. The software then sources the part and provides a part cost and estimated labour cost to replace it. Once this is completed for all parts effected by the accident, the final estimate for the repair is determined.

[0016] However, as noted above, until a new vehicle make / model has been on the road for a substantial amount of time, it is difficult for insurers to know what the cost of repairs may be. Current insurers are making “educated guesses” based on previous models or similar vehicles in order to predict accurate premiums for customers requesting insurance coverage on new makes or models of vehicles. However, when a new make / model of a vehicle is released, the model of the vehicle is typically available in the estimation modeling software. The proposed system and method combine the extensive data collected and retained by insurance companies with estimation modeling software to provide more accurate premium estimates for customers, particularly those requesting insurance for new vehicle makes / models.

[0017] Insurance companies have an extensive amount of information on historical claims, including but not limited to, information about the customer who made the claims (age, car, annual mileage, etc.), and also all the information about the historical claim, (for example, the cost of repair, was it a total loss or not, point of impact on car, details of everything that was repaired on the car if it was repairable, etc.). Based on historical claim information and optionally information about a current client, basic statistics are used to determine either general accident profiles or client specific accident profiles. For the purposes of this disclosure, the mostly likely types of accidents, with their associated likelihood of occurring, is called an accident profile 208.General Accident Profiles

[0018] The preferred embodiment shown in figure 3 uses general accident profiles 308 which are not linked to the types of accidents that a specific driver is expected to encounter. The generic profile is based on all accident events that all drivers are likely to suffer. It’s the “overall weighted average” of all accident types and their frequency measured against the model of the new subject vehicle. A general accident profiles, for example, could be “in rural areas, 20% of impacts are front impacts, 40% are side impact, etc.” For example, the accident profile 308 could be determined based on the common accident types for a particular class of vehicles or based on the most common accidents for the same make and model ofprevious years. This general accident profile can be created each time a new car is coming to the market.Customer Specific Accident Profiles

[0019] In another embodiment, shown in figure 2, information specific to a particular customer 203 is used as input to determine a customer-specific accident profile. In this approach, the customer information, such as where they live, their characteristics, their car, etc., along with historical claim information, is used to determine a list of the typical accident types someone with the customer’s characteristics will likely experience. For example, if the customer is a 35-year-old female that lives in a suburban area, the data within the database can be used as input into an algorithm to determine an accident profile. An accident profile determines the likelihood of different types of accidents corresponding to a customer profile. In the example customer above, the 35-year-old female living in a suburban area, may have an accident type profile that includes 45% rear end collisions, 25%head on collisions, 20%side-impact collisions and 10% sideswipes.

[0020] In yet a further embodiment, accident profiles can be created on any relevant variable, such as urban vs rural settings, age categories of drivers, etc. It will be appreciated by a person skilled in the art that the accident profile 308 or 208 can be determined based on a wide variety of characteristics and should not be limited to the specific examples disclosed herein.Estimating Costs of Repairs

[0021] The overall estimation system for using general accident profiles 300 is shown in figure 3. When a customer requests insurance coverage on a new make / model of car, the data within the insurance database 206 is used as input to create a general accident profile 308. The database 206 contains a plethora of information which is used to estimate insurance premiums for existing vehicles. For example, the database 206 includes data regarding types of accidents, the area in which the accident took place, age of the driver, gender of the driver, make and model of the vehicle in the accident, etc. This data is used to create the general accident profile 308 based on similar vehicles (for example, the previous model).

[0022] In the customer-specific system 200 shown in figure 2, when a customer requests insurance coverage on a new make / model of a car at step 202, all of the customer information is entered 204 into the estimation system 200 and risk factors are extracted from the customer information For example, the customer age, gender, and the area in which they live and the make / class of car they will be driving may be relevant risk factors to consider when calculating their premium. Note that this customer information could alternatively be pulled from existing databases. This information is used in combination with the data within the insurance database 206 to create a customer specific accident profile 208.

[0023] Once the accident profile (either general or customer specific) is determined, for each accident type in the accident profile, the insurance database includes deformation measurements or photographs 204 and damage history of previous accidents from vehicles similar in build to the new vehicle. For example, in some cases, the accident data from same model of car from a previous year could give insight into the predicated deformation 204 and damage of the new vehicle in each of the identified accident types. While in the preferred embodiment, estimated deformation is used to predict the damage that will be sustained by a new vehicle, it can be appreciated that other methods of determining damage, such as but not limited to accident modeling could be used. In a preferred embodiment, damage is predicted through digitally simulating accidents within the accident profile. In this embodiment, a 3D model is used to digitally model crashes within the accident profile of the new car model.Using this method, it is possible to model hundreds or thousands of the same accident type to help estimate what part of the car are likely to be damaged in each accident type. By modeling the accidents digitally, deformation can be estimated without relying on real -world claims data. The modeled deformation can optionally be compared to deformation estimates based on previous or similar models. In one embodiment, both deformation estimates based on previous or similar models and deformation estimates based on digital modeling are used to determine likely deformations for each accident type in the accident profile.

[0024] Using the deformation 204 as input into the estimation modeling software 210. The estimation modeling software 210 uses the model of the car and part configuration 212 in combination with the deformation estimates 204 to determine which parts would be damaged 214 in each accident type. This can be done manually or using computer methods, preferably attended and / or unattended digital models. The estimation modeling software 210 is used tosource and price the replacement parts required 216 and estimates the cost of labour 218. This results in an approximate cost to repair per accident type 220 from which a weighted average cost per accident 222 can be calculated. This can then be used as an input (with or without other risk factors) into known algorithms for determining insurance premiums 224.

[0025] More preferably, however, the costs of each accident type are used to create relative ranks of the new make / model compared to previous vehicles. Once the cost of repairs is known for each accident type within an accident profile, the cost can be used to develop a rating factor in models for determining premiums. Models for developing a rating factor could be based on historical claim data, including, for example, historical cost to repair similar vehicles. For example, if the average severity between an older model and the expected damage of the new model shows that the new model is 25 percent more expensive to repair, premiums can be adjusted accordingly. In another embodiment, the new make / model is compared to other vehicles with similar generic accident profiles and is assigned a relativity rating. For example, if the 2024 model has a cost to repair index of 1.0, we can use the generic accident profile in combination with the estimator modeling software to determine the cost to repair for a 2025 model at 1.14. This would then optionally be further adjusted based on customer-specific factors. Furthermore, if we consider the example where accident profiles are built based on categories, such as urban and rural, multiple general rankings or relativities can be determined for the new make / model. These are all examples of additional variables that can be created and considered when actuaries determine the price of a customer’s premiums. This method of estimating premiums is more accurate and realistic than current methods particularly for new make / models of vehicles.

[0026] It can be appreciated that as a new car model is on the road and is involved in accidents, that actual accident data 226 and actual cost of repair 228 can be fed back in the system as input to adjust the estimated cost per accident type accordingly. By using the actual damage, deformation & repair cost data, the estimating system can be adjusted to more accurately determine how much damage will occur to a particular part installed at a particular location. This feedback is used in future crash simulations to improve the damage estimations of the system over time based on real life experiences.

[0027] Once actual accidents occur with the new vehicles, the estimation modeling software can also be adjusted to reflect actual labour time and actual part cost. Furthermore,as more actual accidents occur, this data can be compared to the accident profile (general or specific) and methods to determine the accident profile can be adjusted to better reflect the real-life data.

[0028] In another embodiment, system 400 shown in figure 4, the estimated characteristics of the each of the accident types in the accident profile are used as input into a vehicle crash simulation software 402. This gives greater insight and accuracy into the type of damage sustained by a new vehicle in various accident types. Once the accident type is modeled, the parts sustaining damage is determined and the cost to repair is estimated using the estimation modeling software 210. In this embodiment, using customer profile information 203 is optional. Furthermore, real accident characteristics 226 and real costs to repair 228 can be fed back into the system 400 to provide more accurate costs of each accident type.Although the invention has been described with reference to certain specific embodiments, various modifications thereof will be apparent to those skilled in the art without departing from the spirit and scope of the invention as outlined in the claims appended hereto. The entire disclosures of all references recited above are incorporated herein by reference.

Claims

What is claimed is:

1. A method for determining insurance premiums for a vehicle comprising the steps of: a processor accessing a database containing accident claim data; the processor calculating a predicted accident profile; the predicted accident profile comprising types of accidents; inputting the accident profile into estimation modeling software to determine an approximate repair cost factor based on determined accident type; the processor using the approximate repair cost factor to determine insurance premiums for the vehicle.

2. The method of claim 1 wherein the vehicle is a new make or model.

3. The method of claim 2 wherein the accident profile further comprises weighting factors associated with each of the determined accident types; the weighting factors associated with each of the accident types being used to determine the repair cost factor by calculating a weighted average of the cost of repair of each accident type in the accident profile.

4. The method of claim 3 wherein the estimation modeling software uses deformation estimates obtained from the database to determine: parts likely to require replacement, cost of parts likely to require replacement, and approximate cost of labour to replace parts likely to require replacement; which are used to calculate the approximate cost of repair for each accident type.

5. The method of claim 3 wherein the estimation modeling software uses deformation estimates obtained from generating digital simulations of each accident type in the accident profile to determine: parts likely to require replacement, cost of parts likely to require replacement, and approximate cost of labour to replace parts likely to require replacement;which are used to calculate the approximate cost of repair for each accident type.

6. The method of claim 3 wherein the estimation modeling software uses deformation estimates obtained from generating digital simulations of each accident type in the accident profde in combination with deformation estimates obtained from the database to determine: parts likely to require replacement, cost of parts likely to require replacement, and approximate cost of labour to replace parts likely to require replacement; which are used to calculate the approximate cost of repair for each accident type.

7. A system for determining insurance premiums for a vehicle comprising: a processor; an estimation model; and a database containing accident claim data; the processor configured to access the database containing accident claim data and use the accident claim data to predict an accident profile; the accident profile comprising types of accidents; the estimation model configured to accept the accident profile as input to determine an approximate repair cost factor based on determined accident type; the processor then configured to use the approximate repair cost factor to determine insurance premiums for the vehicle.

8. The system of claim 7 wherein the vehicle is a new make or model.

9. The method of claim 8 wherein the accident profile further comprises weighting factors associated with each of the determined accident types; the weighting factors associated with each of the accident types being used to determine the repair cost factor by calculating a weighted average of the cost of repair of each accident type in the accident profile.

10. The system of claim 8 wherein the estimation model uses deformation estimates obtained from the database to determine:parts likely to require replacement, cost of parts likely to require replacement, and approximate cost of labour to replace parts likely to require replacement; which are used to calculate the approximate cost of repair for each accident type.

11. The system of claim 8 wherein the estimation model uses deformation estimates obtained from generating digital simulations of each accident type in the accident profde to determine: parts likely to require replacement, cost of parts likely to require replacement, and approximate cost of labour to replace parts likely to require replacement; which are used to calculate the approximate cost of repair for each accident type.

12. The system of claim 8 wherein the estimation model uses deformation estimates obtained from generating digital simulations of each accident type in the accident profde in combination with deformation estimates obtained from the database to determine: parts likely to require replacement, cost of parts likely to require replacement, and approximate cost of labour to replace parts likely to require replacement; which are used to calculate the approximate cost of repair for each accident type.

Images