METHOD AND SYSTEM TO DETERMINE AUTO INSURANCE RISK

• US 20170109827A1
• Filed: 10/15/2015
• Published: 04/20/2017
• Est. Priority Date: 10/15/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: HUANG, He Yuan, DUAN, Ning, WANG, Zhi Hu, LI, Kai
• Granted Patent: US 10,891,693 B2
• Time in Patent Office: Days
• CPC Class Codes: G06Q 40/08 Insurance
• CPC Class Codes: G07C 5/008 communicating information to a remotely located station transmission systems for measured values G08C
• CPC Class Codes: G07C 5/0808 Diagnosing performance data testing of vehicles G01M17/00; testing of electrical installation on vehicles G01R31/005

Abstract

A method for evaluating fraudulent data in a Usage Based Insurance (UBI) system, includes retrieving trip data for a driver from a database. A processor on a computer determines tough context incidents in the trip data. Driving behavior of the driver during said tough context incidents is compared with driving behavior of other drivers during similar tough context incidents. The trip data is identified as potentially fraudulent if the driver'"'"'s driving behavior is better by a predetermined amount compared to the other drivers'"'"' driving behavior.

Claims

1. A method, comprising:

receiving trip data for a driver;

determining, using a processor on a computer, tough context incidents in said trip data;

comparing driving behavior of said driver during said tough context incidents with driving behavior of other drivers during similar tough context incidents; and

identifying said trip data as potentially fraudulent if said driver'"'"'s driving behavior is better by a predetermined amount, compared to said other drivers'"'"' driving behavior.

2. The method of claim 1, wherein said trip data comprises data transmitted to an insurance processing center under a Usage Based Insurance (UBI) model.

3. The method of claim 1, wherein:

said trip data comprises data transmitted from a risk meter in a vehicle being driven by said driver,

wherein said transmitted data is stored in a memory device, and

wherein said receiving trip data comprises retrieving said data from said memory device.

4. The method of claim 1, wherein said comparing comprises a statistical comparison of normalized driving behaviors.

5. The method of claim 1, wherein said tough context incidents are determined using contextual data, said contextual data comprising at least one of data outside a control of said driver and data concerning conditions or events outside a vehicle being driven by said driver.

6. The method of claim 1, wherein said tough context incidents comprise one or more of:

weather conditions designated as hazardous;

traffic designated as congested;

driving during traffic periods designated as more hazardous; and

driving for periods designated as fatigue driving.

7. The method of claim 3, wherein said tough context incidents are determined based on parameter values for permutation combinations of different driving contexts, said driving contexts comprising at least road type, speed pattern, and traffic period.

8. The method of claim 1, wherein said trip comprises a first trip, said method further comprising using said trip data to develop a single trip analysis for said first trip, said trip analysis comprising a matrix listing of driving behaviors by said driver along points of said first trip along with driving context data along segments of said first trip.

9. The method of claim 8, further comprising:

using available trip data of other trips by said driver and available trip data of trips of other drivers to develop an overall profile analysis; and

using said overall profiling analysis to determine risk hot spots along a path of said first trip and to determine benchmarks along said path of said first trip, said risk hot spots comprising locations along said path at which driving behaviors of drivers indicate driving hazards, said benchmarks comprising permutation combinations of different driving contexts including at least road type, speed pattern, and traffic period at which driving behaviors indicate driving hazards, said risk hot spots and said benchmarks indicating driving hazards beyond a preset threshold comprising said tough context incidents.

10. The method of claim 9, further comprising:

using said risk hot spots and said benchmarks to develop a personal profiling analysis for said driver;

calculating risk hot spot and benchmark data statistics for all drivers;

determining whether said personal profiling analysis for said driver demonstrates superior driving behavior compared to a predetermined threshold value of said data statistics for all drivers;

receiving any available constraint data that might explain any of said driver'"'"'s superior driving behavior; and

outputting said driver'"'"'s personal profiling analysis as potentially a fraudster if said driver demonstrates superior driving behavior and there is insufficient constraint data to explain such superior driving behavior.

11. The method of claim 1, as embodied in a set of computer-readable instructions stored on a non-transitory storage device.

12. The method of claim 11, wherein said non-transitory storage device comprises one of:

a memory in a computer storing instructions currently being executed by said computer;

a memory on said computer storing instructions to selectively be executed;

a standalone memory device capable of being inserted into a port of a computer to upload said computer-readable instructions;

a memory in a network server providing a service to download said computer-readable instructions onto another computer accessible to said network server;

a memory in a network server providing a service to perform said method; and

a memory in a cloud service providing a service to perform said method.

13. The method of claim 1, wherein said trip data comprises trip data for repetitive trips made by said driver.

14. An apparatus, comprising:

a processor; and

a memory device storing a set of computer-readable instructions to perform a method of evaluating driver data as potentially fraudulent, said method comprising;

receiving trip data for a driver;

determining, using a processor on a computer, tough context incidents in said trip data;

comparing driving behavior of said driver during said tough context incidents with driving behavior of other drivers during similar tough context incidents; and

identifying said trip data as potentially fraudulent if said driver'"'"'s driving behavior is statistically better by a predetermined amount, compared to said other drivers'"'"' driving behavior.

15. The apparatus of claim 14, as comprising a computer in an insurance processing center receiving driver data under a Usage-Based Insurance (UBI) model.

16. The apparatus of claim 14, as comprising a computer participating in a cloud service providing said method as a service.

17. A method of detecting potential fraud under a Usage-Based Insurance (UBI) model, said method comprising:

collecting driver trip data from drivers covered under the UBI model and storing said driver trip data in a memory device, said driver trip data having associated therewith contextual data concerning said trip;

selecting a driver considered as an excellent driver;

retrieving data from said memory device for a trip made by said excellent driver;

determining incidents of tough context driving during said trip, using said contextual data;

comparing said excellent driver'"'"'s driving behavior during said tough context driving incidents with driving behavior of other drivers during similar tough context driving incidents;

determining whether said excellent driver'"'"'s driving behavior during said tough context driving incidents is better by a predetermined amount from said other drivers'"'"' driving behavior; and

identifying said excellent driver as a potential fraudster if it is determined that said excellent driver'"'"'s driving behavior exceeds said predetermined amount.

18. The method of claim 17, further comprising:

if said excellent driver'"'"'s driving behavior exceeds said predetermined amount, retrieving constraint data concerning said driving behavior, said constraint data comprising driving behavior of other drivers under a same spatiotemporal condition and a same tough driving context incident;

comparing said excellent driver'"'"'s driving behavior with driving behavior of said constraint data; and

determining whether said constraint data provides sufficient explanation for said excellent driver'"'"'s better driving behavior.

19. The method of claim 17, wherein said comparing comprises a statistical comparison of normalized driving behaviors.

20. The method of claim 17, as embodied in a set of computer-readable instructions stored on a non-transitory storage device.