CEO Dynamotion Medical Henderson, Nevada, United States
Abstract Details:
Background: There are multiple means of diagnosing and managing cognitive conditions such as concussions and traumatic brain injury (TBI) from simple questioning and observation to radiological imaging and brain MRI. Yet there is not a uniform testing program that includes both a means of immediate assessment on the field and comprehensive clinical management using a validated instrument. This protocol uses various testing modalities based upon tactile edge orientation processing (TEOP), motion signatures, and machine learning as a rapid test and classification of a cognitive condition. The first assessment of this was performed using cognitive impairment with blood alcohol content (BAC) of 0.00 to 0.09%. A portion of this work was granted in US Patent #11,751,799 B1 September 12, 2023. The machine learning innovation described herein was never previously published. This was done to show proof of concept for a larger, ongoing study to evaluate the potential development of cognitive conditions in retired NFL players. Purpose: To identify a rapid cognitive test to properly classify the presence and severity of a cognitive condition using tactile edge orientation processing, motion signatures, and machine learning. Methods: Random subjects (60% men; 40% women) with varied ages participated. Each subject was included in the control (sober) and experimental (intoxication up to 0.09% BAC) groups. This resulted in a sample size of 20 divided between train and test groups for the machine learning algorithm. The subjects were tested sober and after being given controlled amounts of alcohol. BAC was measured prior to running the test. Two motion tests were performed, 1) motion of wrist when telling time and 2) sliding their ankle along the tibia of the opposite leg while in the seated position. A MEMS sensor including an accelerometer and gyroscope was placed on each subject’s wrist / leg to record a motion signature relative to gravity during the test. The MEMS sensor was wirelessly connected via BLE to a laptop computer where motion signatures in the X, Y, and Z directions for both acceleration and angular rotation were collected. The test was performed at the patient’s own pace. The data was displayed visually like an EKG in graphs and machine learning classification algorithms were applied to the motion signature data from both tests. Model accuracy was then evaluated. Results: Machine learning properly classified 80% of the cases but there was more to the story. Careful review of the data showed men lost more motor control than women with intoxication. Hence the machine learning properly classified all sober subjects and men intoxicated, but not women intoxicated. This correlated well with another study by the authors that achieved 96% accuracy using the same ML models and specific motions for shoulder injury (rotator cuff tear) diagnosis. Conclusions: The utilization of machine learning and TEOP has the potential to properly classify a cognitive condition such as sober versus intoxicated for men with additional testing needed to validate this approach with women (possibly different TEOP tests) who demonstrate superior motor control while cognitively impaired. PRACTICAL APPLICATION: Concussions and TBI often occur in athletes playing popular sports, police, and soldiers in the field. This approach may be a fast, standardized method of diagnosis and monitoring of these cognitive conditions to allow patients to safely return to play and work.
