DATA COLLECTION AND ANALYSIS FOR THE PURPOSE OF ENABLING AI TRIAGE OF VEHICLE ACCIDENTS:
CLOUD COMPUTING & BIG DATA
Author: Timothy Kasprzak
Status: Project Concept
The National Highway Traffic Safety Administration has established Event Data Recorders (EDRs) or “black boxes” to capture data in a traffic accident. In 2013, 96% of new cars sold in the United States were equipped with a black box and on September 1, 2014 all new vehicles must have a black box installed (Federal Register Vol. 77, No. 240, 49 CFR Part 571 Event Data Recorders).
These EDRs the data must include (Federal Register Vol. 77, No. 154, 49 CFR Part 563 Event Data Recorders):
• The forward and lateral crash force.
• The crash event duration.
• Indicated vehicle speed.
• Accelerator position.
• Engine rpm.
• Brake application and antilock brake activation.
• Steering wheel angle.
• Stability control engagement.
• Vehicle roll angle, in case of a rollover.
• Number of times the vehicle has been started.
• Driver and front-passenger safety belt engagement, and pretensioner or force limiter engagement.
• Air bag deployment, speed, and faults for all air bags.
• Front seat positions.
• Occupant size.
• Number of crashes (one or more impacts during the final crash event).
I propose that in a comparable way to tasking AI to analyze vast volumes of medical data to generate predictive analytics, we apply AI to vehicle accident data and identify any correlations with passenger injuries. I envision a three-step process:
First, anonymized data collection with the above data points and additional vehicle information such as make, model, year, environmental conditions. Data collection will also include correlative passenger-patient data, such demographics and final injury diagnosis as based on grading criteria of the American Association for the Surgery of Trauma.
Second, data analysis by AI to identify associations between the vehicle data and type/severity of patient injuries.
Third, utilize the validated associations to enable the vehicles to triage probability of types of injury to first responders and medical personnel by onboard signaling technology.