Patent Issued for Collision analysis platform using machine learning to reduce generation of false collision outputs (USPTO 11562603): Allstate Insurance Company

2023 FEB 15 (NewsRx) -- By a News Reporter-Staff News Editor at Insurance Daily News -- According to news reporting originating from Alexandria, Virginia, by NewsRx journalists, a patent by the inventors Deram, Jeremy (Northbrook, IL, US), Schmitt, Kyle Patrick (Chicago, IL, US), Tammali, Venu Madhav (Chicago, IL, US), filed on June 26, 2020, was published online on January 24, 2023.

The assignee for this patent, patent number 11562603, is Allstate Insurance Company (Northbrook, Illinois, United States).

Reporters obtained the following quote from the background information supplied by the inventors: “Aspects of the disclosure relate to enhanced processing systems for executing machine learning algorithms and automatically determining whether or not a collision occurred. Many organizations and individuals use sensor data to determine whether or not a collision occurred. In many instances, however, these determinations may result in false positive and/or false negative results. In addition to the inaccuracies caused by such results, unnecessary resources may be expended or deployed in response to a false positive determination. Similarly, resources may be wrongfully conserved or withheld in response to a false negative determination.”

In addition to obtaining background information on this patent, NewsRx editors also obtained the inventors’ summary information for this patent: “Aspects of the disclosure provide effective, efficient, scalable, and convenient technical solutions that address and overcome the technical problems associated with real time on the edge automated collision determinations. In accordance with one or more arrangements discussed herein, a computing platform having at least one processor, a communication interface, and memory may receive sensor data. By applying one or more machine learning algorithms to the sensor data, the computing platform may generate an event and an associated confidence of whether or not the event is a collision event. In addition to generating the event and a confidence of whether a collision occurred, the computing platform may: 1) identify the location where the event is triggered, and 2) determine whether or not the location is within a predetermined radius of a false positive collection location. In response to determining that the data collection location is within the predetermined radius, the computing platform may modify the event confidence to indicate that a collision did not occur. In response to determining that the data collection location is not within the predetermined radius, the computing platform may: 1) analyze telematics data included in the sensor data to modulate the confidence of collision score, and 2) compare confidence of collision score to a predetermined collision threshold. In response to determining that the confidence of collision score does not exceed the predetermined collision threshold; the computing platform may modify the collision output to indicate that a collision did not occur.

“In response to determining that the confidence of collision score exceeds the predetermined collision threshold, in one or more instances, the computing platform may analyze angular velocity data included in the sensor data, by comparing the angular velocity data to one or more machine learning datasets corresponding to non-collision events in which a mobile device was dropped, to further modulate confidence of the collision score. The computing platform may compare the confidence of collision score to the predetermined collision threshold. In response to determining that the confidence of collision score does not exceed the predetermined collision threshold; the computing platform may modify the collision output to indicate that a collision did not occur. In response to determining that the confidence of collision score exceeds the predetermined collision threshold, in one or more instances, the computing platform may compute a time difference between trip start time and event time. In response to determining that the time difference is below a predetermined time threshold; the computing platform may modify the collision output to indicate that a collision did not occur.

“In one or more examples, the sensor data may be received from one or more of: a mobile device or vehicle based sensors. In one or more instances, the computing platform may analyze one or more of: a data collection location, telematics data, or angular velocity corresponding to a second mobile device to generate a second collision output, where the sensor data is provided to the computing platform, at least in part, by the second mobile device. The computing platform may determine whether the second collision output indicates that a collision did occur. In response to determining that the second collision output indicates that a collision did not occur, the computing platform may modify the collision output to indicate that a collision did not occur. In response to determining that the second collision output indicates that a collision did occur, the computing platform may affirm the collision output indicating that a collision did occur.

“In one or more examples, the computing platform may send a corroboration request to a mobile device corresponding to the sensor data. The computing platform may receive, from the mobile device, crash confirmation information indicating whether or not a collision occurred, which may be based on user input received at the mobile device indicating whether or not the collision occurred. In response to determining that the crash confirmation information indicates that the collision did not occur, the computing platform may modify the collision output to indicate that a collision did not occur. In response to determining that the crash confirmation information indicates that the collision did occur, the computing platform may affirm the collision output indicating that a collision did occur.

“In one or more instances, in response to generating the collision output indicating that a collision did not occur, the computing platform may compare barometric data included in the sensor data to a predetermined airbag deployment threshold. In response to determining that the barometric data exceeds the predetermined airbag deployment threshold, the computing platform may modify the collision output to indicate that a collision did occur. In one or more examples, in response to not generating a collision output indicating that a collision did not occur, the computing platform may: 1) compare one or more thresholds determined using the telematics data with the predetermined collision threshold, wherein the predetermined collision threshold is derived, using Machine Learning, from the dataset formed by merging historical telematics data and historical claims data; 2) in response to determining that the one or more thresholds are greater than the predetermined collision threshold, modify the collision output to indicate that a collision occurred; and 3) in response to determining that the one or more thresholds are lower than the predetermined collision threshold, affirm the collision output indicating that a collision did not occur.”

The claims supplied by the inventors are:

“1. A computing platform, comprising: at least one processor; a communication interface communicatively coupled to the at least one processor; and memory storing computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: receive sensor data; generate, by applying one or more machine learning algorithms to the sensor data, a collision output indicating whether or not a collision occurred; in response to generating the collision output indicating that a collision occurred: identify a data collection location corresponding to the sensor data; determine whether or not the data collection location is within a predetermined radius of a false positive collection location; in response to determining that the data collection location is within the predetermined radius, modify the collision output to indicate that a collision did not occur; in response to determining that the data collection location is not within the predetermined radius; analyze telematics data included in the sensor data to compute a first likelihood of collision score; compare the first likelihood of collision score to a predetermined collision threshold; in response to determining that the first likelihood of collision score does not exceed the predetermined collision threshold, modify the collision output to indicate that a collision did not occur; and in response to determining that the first likelihood of collision score exceeds the predetermined collision threshold, affirm the collision output indicating that a collision did occur; and based on a determination that the collision output indicates that a collision did occur, send one or more commands to a dispatch computing system directing the dispatch computing system to dispatch a service vehicle to a location of the collision, wherein sending the one or more commands to the dispatch computing system directing the dispatch computing system to dispatch the service vehicle to the location of the collision causes the dispatch computing system to dispatch the service vehicle to the location of the collision.

“2. The computing platform of claim 1, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: analyze angular velocity data included in the sensor data to compute a second likelihood of collision score; compare the second likelihood of collision score to the predetermined collision threshold; in response to determining that the second likelihood of collision score does not exceed the predetermined collision threshold, modify the collision output to indicate that a collision did not occur; and in response to determining that the second likelihood of collision score exceeds the predetermined collision threshold, affirm the collision output indicating that a collision did occur.

“3. The computing platform of claim 2, wherein analyzing the angular velocity data comprises comparing the angular velocity data to one or more machine learning datasets corresponding to non-collision events in which a mobile device was dropped.

“4. The computing platform of claim 1, wherein: the computing platform comprises one or more of: a mobile device that collected the sensor data or a computing system integrated into a vehicle corresponding to the collision output; and the sensor data is received from one or more of: a mobile device or vehicle based sensors.

“5. The computing platform of claim 1, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: analyze one or more of: a data collection location, telematics data, or angular velocity corresponding to a second mobile device to generate a second collision output, wherein the sensor data is provided to the computing platform, at least in part, by the second mobile device; determine whether the second collision output indicates that a collision did occur; in response to determining that the second collision output indicates that a collision did not occur, modify the collision output to indicate that a collision did not occur; and in response to determining that the second collision output indicates that a collision did occur, affirm the collision output indicating that a collision did occur.

“6. The computing platform of claim 1, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: send a corroboration request to a mobile device corresponding to the sensor data; receive, from the mobile device, crash confirmation information indicating whether or not a collision occurred, wherein the crash confirmation information is based on user input received the mobile device indicating whether or not the collision occurred; in response to determining that the crash confirmation information indicates that the collision did not occur, modify the collision output to indicate that a collision did not occur; and in response to determining that the crash confirmation information indicates that the collision did occur, affirm the collision output indicating that a collision did occur.

“7. The computing platform of claim 1, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: in response to not generating a collision output indicating that a collision did not occur: compare barometric data included in the sensor data to a predetermined airbag deployment threshold; and in response to determining that the barometric data exceeds the predetermined airbag deployment threshold, modifying the collision output to indicate that a collision did occur.

“8. The computing platform of claim 1, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: in response to not generating a collision output indicating that a collision did not occur: compare one or more thresholds determined using the telematics data with the predetermined collision threshold, wherein the predetermined collision threshold is derived, using Machine Learning, from the dataset formed by merging historical telematics data and historical claims data; in response to determining that the one or more thresholds are greater than the predetermined collision threshold, modify the collision output to indicate that a collision occurred; and in response to determining that the one or more thresholds are lower than the predetermined collision threshold, affirm the collision output indicating that a collision did not occur.

“9. The computing platform of claim 1, wherein the false positive collection location comprises one or more of: a ski resort, an amusement park, or a body of water.

“10. The computing platform of claim 1, wherein analyzing the telematics data included to compute the first likelihood of collision score comprises comparing the telematics data to one or more machine learning datasets corresponding to a roller coast event, a ski event, or a boat event.

“11. The computing platform of claim 1, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: update, after modifying the collision output, the one or more machine learning algorithms to indicate that a false positive collision determination was made by the one or more machine learning algorithms.

“12. A method comprising: at a computing platform comprising at least one processor, a communication interface, and memory: receiving sensor data; generating, by applying one or more machine learning algorithms to the sensor data, a collision output indicating whether or not a collision occurred; in response to generating the collision output indicating that a collision occurred: analyzing telematics data included in the sensor data to compute a first likelihood of collision score; comparing the first likelihood of collision score to a predetermined collision threshold; in response to determining that the first likelihood of collision score does not exceed the predetermined collision threshold, modifying the collision output to indicate that a collision did not occur, and in response to determining that the first likelihood of collision score exceeds the predetermined collision threshold: identifying a data collection location corresponding to the sensor data, determining whether or not the data collection location is within a predetermined radius of a false positive collection location, in response to determining that the data collection location is within the predetermined radius, modifying the collision output to indicate that a collision did not occur, and in response to determining that the data collection location is not within the predetermined radius, affirming the collision output indicating that a collision did occur; and based on a determination that the collision output indicates that a collision did occur, sending one or more commands to a dispatch computing system directing the dispatch computing system to dispatch a service vehicle to a location of the collision, wherein sending the one or more commands to the dispatch computing system directing the dispatch computing system to dispatch the service vehicle to the location of the collision causes the dispatch computing system to dispatch the service vehicle to the location of the collision.”

There are additional claims. Please visit full patent to read further.

For more information, see this patent: Deram, Jeremy. Collision analysis platform using machine learning to reduce generation of false collision outputs. U.S. Patent Number 11562603, filed June 26, 2020, and published online on January 24, 2023. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(11562603)&db=USPAT&type=ids

(Our reports deliver fact-based news of research and discoveries from around the world.)