Systems and methods for determining likelihood of incident relatedness for diagnostic trouble codes

A non-transitory machine-readable storage medium encoded with instructions executable by a hardware processor of a computing component, the machine-readable storage medium comprising instructions to cause the hardware processor to perform operations comprising:

• receiving a repair diagnostic trouble code (DTC) of a type that is caused by mechanical component malfunction, the DTC specifying diagnostic information related to a component in a vehicle which has experienced a collision event, and the diagnostic information comprising one or more data inputs generated by the component in the vehicle;
• generating a repair estimate for the vehicle for the collision event to include the DTC;
• generating a set of input data comprising the DTC and at least one of:
• the repair estimate, an image of the vehicle following the collision event, and an insurance claim for the collision event;
• applying the set of input data as an input to a trained machine learning model, wherein the trained machine learning model has been trained using historical examples of the set of input data, wherein responsive to the input, the trained machine learning model outputs a likelihood that the DTC was caused by the collision event;
• determining a vehicle make/model of the vehicle;
• determining whether the DTC is present in a list of DTCs that are known to not be caused by collision events by utilizing a lookup function on a database filterable by vehicle makes/models that stores the list of DTCs and filtering the list of DTCs to include only those with a same vehicle make/model as the vehicle;
• determining that the DTC is unrelated to the collision event when the likelihood that the DTC was caused by the collision event is below a preset threshold likelihood based on the trained machine learning model output and the DTC is present in the list of DTCs that are known to not be caused by collision events when utilizing the lookup function on the database;
• when the DTC is determined to be unrelated to the collision event, removing the DTC from the repair estimate;
• and
• outputting the repair estimate to an insurer claims management system.

A method implemented by a server computer, the method comprising:

• receiving a repair diagnostic trouble code (DTC), the DTC specifying diagnostic information related to a component in a vehicle which has experienced a collision event, and the diagnostic information comprising one or more data inputs generated by the component in the vehicle;
• generating a repair estimate for the vehicle for the collision event to include the DTC;
• generating a set of input data comprising the DTC and at least one of:
• the repair estimate, an image of the vehicle following the collision event, and an insurance claim for the collision event;
• applying the set of input data as an input to a trained machine learning model, wherein the trained machine learning model has been trained using historical examples of the set of input data, wherein responsive to the input, the trained machine learning model outputs a likelihood that the DTC was caused by the collision event;
• determining a vehicle make/model of the vehicle;
• determining whether the DTC is present in a list of DTCs that are known to not be caused by collision events by utilizing a lookup function on a database filterable by vehicle makes/models that stores the list of DTCs and filtering the list of DTCs to include only those with a same vehicle make/model as the vehicle;
• determining that the DTC is unrelated to the collision event when the likelihood that the DTC was caused by the collision event is below a preset threshold likelihood based on the trained machine learning model output and the DTC is present in the list of DTCs that are known to not be caused by collision events when utilizing the lookup function on the database;
• when the DTC is determined to be unrelated to the collision event, removing the DTC from the repair estimate;
• and
• outputting the repair estimate to an insurer claims management system.