Machine-learning method of loss prediction using incident information

1 . A computing system for machine learning prediction of damage loss for insurance claims, comprising:

a network communication interface communicatively coupled to a data network;

one or more processors communicatively coupled to the network communication interface; and

a memory, communicatively coupled to the one or more processors, storing instructions that, when executed by the one or more processors, cause the computing system to perform operations comprising:

training one or more models to make total loss predictions for vehicles involved in vehicle collisions, based on historical vehicle incident data obtained from prior vehicle incidents, the training including (i) determining, from prior vehicle incidents, discrepancies in historical determinations between predicted damage to vehicles involved in vehicles incidents with actual payout for those vehicle incidents; and (ii) tuning the one or more models based on the determined discrepancies;

providing, over the data network, to a computing device of a user involved in a vehicle incident, an interface including a three-dimensional representation of a vehicle, the three-dimensional representation being specific to a set of characteristics of a vehicle involved in the vehicle incident, the interface enabling the user to rotate the three-dimensional representation about one or more axes;

enabling the user to specify vehicle incident data for the vehicle incident by interacting with the three-dimensional representation and marking areas of the three-dimensional representation to indicate location of damage to the vehicle;

based at least in part on the vehicle incident data, generating a total loss prediction using the one or more models, the total loss prediction indicating whether the vehicle is repairable or totaled relative to a value of the vehicle;

based on the total loss prediction indicating that the vehicle is repairable at an estimated repair cost, selecting a service provider for the vehicle, and contacting and scheduling, over the data network, repair with the selected service provider, the service provider being selected to match the total loss prediction; and

based on the total loss prediction indicating that the vehicle is totaled, coordinating a tow service to tow the vehicle to a scrap yard or a salvage yard.

2 . The computing system of claim 1 , wherein the vehicle incident data is received via the claim process in which a plurality of individuals involved in the vehicle incident provide contextual information corresponding to the vehicle incident, and wherein generating the total loss prediction is further based on the contextual information.

3 . The computing system of claim 2 , wherein the vehicle incident data is further provided by one or more additional users interacting with the three-dimensional representation of the vehicle.

4 . The computing system of claim 2 , wherein the operations include:

receiving image data from the user, the image data indicating damage to the vehicle; and

wherein generating the total loss prediction is based at least in part on the image data.

5 . The computing system of claim 4 , wherein the operations include guiding the user in capturing images of the vehicle using the computing device of the user.

6 . The computing system of claim 1 , wherein generating includes determining a state of the vehicle, the state indicating whether the vehicle is repairable, totaled, or salvageable.

7 . The computing system of claim 6 , wherein selecting is based at least in part on the state of the vehicle.

8 . The computing system of claim 1 , wherein the operations include receiving the vehicle incident data from the computing device of the user over one or more sessions.

9 . The computing system of claim 8 , wherein at least a portion of the vehicle incident data is received while the user is on a call with a caller representative.

10 . A non-transitory computer readable medium storing instructions that, when executed by one or more processors of a computing system for machine learning prediction of damage loss for insurance claims, cause the computing system to perform operations that include:

training one or more models to make total loss predictions for vehicles involved in vehicle collisions, based on historical vehicle incident data obtained from prior vehicle incidents, the training including (i) determining, from prior vehicle incidents, discrepancies in historical determinations between predicted damage to vehicles involved in vehicles incidents with actual payout for those vehicle incidents; and (ii) tuning the one or more models based on the determined discrepancies;

providing, to a computing device of a user involved in a vehicle incident, an interface including a three-dimensional representation of a vehicle, the three-dimensional representation being specific to a set of characteristics of a vehicle involved in the vehicle incident, the interface enabling the user to rotate the three-dimensional representation about one or more axes;

enabling the user to specify vehicle incident data for the vehicle incident by interacting with the three-dimensional representation and marking areas of the three-dimensional representation to indicate location of damage to the vehicle;

based at least in part on the vehicle incident data, generating a total loss prediction using the one or more models, the total loss prediction indicating whether the vehicle is repairable or totaled relative to a value of the vehicle;

based on the total loss prediction indicating that the vehicle is repairable at an estimated repair cost, selecting a service provider for the vehicle, and contacting and scheduling repair with the selected service provider, the service provider being selected to match the total loss prediction; and

based on the total loss prediction indicating that the vehicle is totaled, coordinating a tow service to tow the vehicle to a scrap yard or a salvage yard.

11 . The non-transitory computer readable medium of claim 10 , wherein the vehicle incident data is received via a claim process in which a plurality of individuals involved in the vehicle incident provide contextual information corresponding to the vehicle incident, and wherein generating the total loss prediction is further based on the contextual information.

12 . The non-transitory computer readable medium of claim 11 , wherein the vehicle incident data is further provided by one or more additional users interacting with the three-dimensional representation of the vehicle.

13 . The non-transitory computer readable medium of claim 11 , wherein the operations include:

guiding the user in capturing images of the vehicle using the computing device of the user, and wherein generating the total loss prediction is based at least in part on image data of the captured images.

14 . A machine-learning method for machine learning prediction of damage loss for insurance claims, the method being performed by one or more processors and comprising:

training one or more models to make total loss predictions for vehicles involved in vehicle collisions, based on historical vehicle incident data obtained from prior vehicle incidents, the training including (i) determining, from prior vehicle incidents, discrepancies in historical determinations between predicted damage to vehicles involved in vehicles incidents with actual payout for those vehicle incidents; and (ii) tuning the one or more models based on the determined discrepancies;

providing, to a computing device of a user involved in a vehicle incident, an interface including a three-dimensional representation of a vehicle, the three-dimensional representation being specific to a set of characteristics of a vehicle involved in the vehicle incident, the interface enabling the user to rotate the three-dimensional representation about one or more axes;

enabling the user to specify vehicle incident data for the vehicle incident by interacting with the three-dimensional representation and marking areas of the three-dimensional representation to indicate location of damage to the vehicle;

based at least in part on the vehicle incident data, generating a total loss prediction using the one or more models, the total loss prediction indicating whether the vehicle is repairable or totaled relative to a value of the vehicle;

based on the total loss prediction indicating that the vehicle is repairable at an estimated repair cost, selecting a service provider for the vehicle, and contacting and scheduling repair with the selected service provider, the service provider being selected to match the total loss prediction; and

based on the total loss prediction indicating that the vehicle is totaled, coordinating a tow service to tow the vehicle to a scrap yard or a salvage yard.

15 . The method of claim 14 , wherein the vehicle incident data is received via a claim process in which a plurality of individuals involved in the vehicle incident provide contextual information corresponding to the vehicle incident, and wherein generating the total loss prediction is further based on the contextual information.