Current Automatic Collision Notification Systems (ACN) utilize deltaV as a simple predictor of injury likelihood. When considered independently, this single variable provides a general indication of injury potential, yet it lacks specificity to adequately distinguish between injured and uninjured occupants in many cases. However, when additional crash attributes are considered in conjunction with deltav, the accuracy of injury predictions greatly improves. The following paper presents two crash models of varied complexity and compares their predictive ability with predictions based on deltaV alone.
Within this paper, regression models are presented which relate occupant, vehicle and impact characteristics to two different injury outcome variables. These are Maximum Abbreviated Injury Scale Level (MAIS) and occupant Injury Severity Score (ISS). The accuracy of proposed models are evaluated using National Automotive Sampling System/ Crashworthiness Data System (NASS/CDS) and Crash Injury Research and Engineering Network (CIREN) case data.
Cumulatively, the positive prediction rate of models identifying the likelihood of MAIS3 and higher injuries is 74.2%. Regression models which predict ISS on a continuous scale correctly identify injured occupants with a sensitivity of 86.1%. The predictive accuracy of each model presented is compared with deltaV alone to support the need for additional model variables for use in future ACN systems.