Model Development and Translation of a Virtual Reality Military Operational Neuropsychological Assessment (VRMONA)
Abstract
Accurate assessment of the extent of injury is critical to appropriate treatment and recovery. However, in the multidomain operational environment, access to co-located medical resources will often be limited or non-existent, so small teams will have to conduct assessment, triage, and treatment in far-forward locations without support from higher echelons. Small team leaders and combat medics need portable, easy to use, and validated methods for assessing cognitive status. Unfortunately, current neuropsychological assessment methods lack portability, require specialized training, and are too cumbersome and time-consuming for use in far-forward military contexts. Thus, there is a critical need to develop alternative automated approaches to neurocognitive assessment that can provide valid determination of injury and return-to-duty status in minimal time. We propose to address this need by capitalizing on recent advances in the fields of artificial intelligence (AI), machine learning, and computational neuroscience to develop a method to simultaneously assess multiple abilities during a brief virtual reality (VR) game-like scenario. Our proposed project, Model Development and Translation of a Virtual Reality Military Operational Neuropsychological Assessment (VRMONA), addresses the FY22 TBIPHRP Focus Area of Prevent and Assess, which focuses on validation of objective tools/methods for assessing and real-time status monitoring of psychological health conditions and/or TBI. To address this critical need, we propose to carry out a 36-month project that involves developing a game-like VR combat scenario that requires the examinee to interact with various characters and solve problems while using a lightweight, commercially available, portable headset system. The game will only last for a few minutes but will continuously record multiple streams of performance data simultaneously. The goal is to extract multiple streams of performance data from the game that will measure neurocognitive performance. Our team includes two computer engineers with extensive experience programming such VR simulations within AI environments. Using an AI technique known as deep neural network (DNN) learning, we will train the system to identify multiple cognitive abilities known to be impacted after mTBI and other neurological conditions. This will be done by having 400 individuals with mTBI and other neurologic disorders, and 200 healthy individuals complete the VR program as well as a traditional comprehensive neuropsychological test battery. Machine learning will be used to extract and combine multiple streams of sensor data to determine the primary cognitive domain scores. To begin with, we will focus mostly on four primary neurocognitive domains that are particularly sensitive to the effects of mTBIs including: (1) attention, (2) processing speed, (3) memory, and (4) executive function (though we plan to expand these in later work). Once the DNN has been trained to identify these domains, we will test the DNN’s ability to predict performance in a new sample of 300 individuals with mTBI and other neurologic conditions and 100 healthy individuals, similar to the training phase. This validation process will allow us to determine the effectiveness of the program at predicting actual neuropsychological performance and accurately identifying people with mTBI in an independent sample. Once validated, future work will test the program out with active military personnel. Within 3 years, we will have a fully functioning prototype system for measuring neuropsychological performance that is based on an individual’s interaction in a VR environment combined with real-time assessment of performance using AI. This is a completely novel approach to identifying brain injury and cognitive impairment and has the potential to revolutionize the field of neuropsychological assessment. Whereas traditional neuropsychological testing involves admini
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 04, 2024
- Source ID
- HT94252310877
Entities
People
- William D. Killgore
Organizations
- United States Army
- University of Arizona