AutoTriage: a feasibility assessment of using in-vehicle sensors to predict traumatic brain injury in road traffic collisions
File(s)
Author(s)
Baker, Claire
Type
Thesis
Abstract
Road traffic collisions (RTCs) are a leading global cause of mortality and morbidity, including for traumatic brain injury (TBI). Injuries sustained often require rapid medical care. TBI patients transferred directly to neurosurgical facilities have improved outcomes. Identifying high risk TBI patients to the emergency medical services (EMS) in real-time via an “AutoTriage” system could improve patient outcomes. Automatic Collision Notification (ACN) systems exist in many modern vehicles. They detect collisions and notify EMS of collision location. Advanced ACN additionally indicate whole-body injury severity, but not brain injury information. In-depth data was used to determine the current TBI severity and pathology collision landscape in Great Britain using a novel free-text algorithm. TBI risk was related to kinematic factors (delta-V, including direction) for vehicle occupants and vulnerable road users (VRUs). For car occupants, individual characteristics (such as seatbelt status, medical conditions, intoxication status and age) were included to improve model prediction. This thesis demonstrates that in-vehicle sensors can predict occupant TBI outcomes from in-depth data, using variables that can be captured by existing sensors. Data and existing sensors are limited for VRUs. Pedestrian-car collisions were reconstructed to validate a virtual modelling approach. A large, simulated database was created. Similar models were created from virtual sensor data to predict injury risk using head kinematics. This demonstrated that VRU data can be augmented effectively using multibody modelling, showing future promise of a VRU-specific AutoTriage system. This timely thesis’ methods and contents fit under the fifth pillar of the UN’s Decade of Action for Road Safety 2021-30: post-collision care. EuroNCAP shares an aligned 2030 vision, which includes post-collision care in vehicle safety rating and incorporates computational modelling. Predicting TBI in RTCs using in-vehicle sensors has the potential to improve post-collision response for TBI casualties, adopting existing technology to benefit society in the imminent future.
Version
Open Access
Date Issued
2023-02
Date Awarded
2023-06
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Ghajari, Mazdak
Sharp, David
Martin, Phil
Wilson, Mark
Sponsor
Transport Research Laboratory (Great Britain)
Engineering and Physical Sciences Research Council
Grant Number
EP/L016737/1
Publisher Department
Dyson School of Design Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)