Mar 11 2020
Better awareness of concussion risks in young athletes has driven scientists to utilize a wide range of head impact sensors to quantify the severity and frequency of impacts during sports.
Children’s Hospital of Philadelphia (CHOP) performed a new study that demonstrates that these head sensors are capable of recording a huge number of false-positive effects at the time of actual gameplay. The study performed by the CHOP group emphasizes that an additional step to confirm the sensor data through a video is required for research and to apply this data in injury prevention strategies to ensure the safety of players.
The study results were published online in March by the American Journal of Sports Medicine.
Around one in five high-school athletes who play a contact sport like American football, lacrosse, and soccer experience a concussion every year. To infer the magnitude, direction, and frequency of head impacts sustained by the athletes, an extensive range of sensors has been designed to gather head impact biomechanics data, such as skull caps, instrumented helmets, headbands, skin patches, and mouthguards.
But false negatives and false positives are likely to occur if data is gathered during gameplay instead in a controlled laboratory setting. In this analysis, CHOP scientists utilized the data gathered from headband-based head impact sensors worn by both female and male soccer players to find out the level of false positives within the data and whether video confirmation enhanced the data quality.
Head impact sensors are a readily accessible tool for studying the mechanics of head impacts. However, in order for researchers to have reliable data to analyze, they first need to verify whether sensor-recorded events are actual head impacts using either video- or observer-confirmation.
Declan Patton, Study Lead Author and Research Associate, Center for Injury Research and Prevention, Children’s Hospital of Philadelphia
In this research, scientists attached headband-fitted impact sensors on 72 junior varsity and high school varsity soccer players (comprising 49 males and 23 females) during 41 games conducted over two seasons to record sensor-recorded events at the time of competition. Every single game was video-recorded. The researchers examined the video to measure the percentage of events captured by the sensors that matched with an observable head impact event.
The events recorded by the sensor and verified by video were also compared with the manufacturer filtering algorithm designed to remove the false-positives.
The sensors were able to record 9,503 events during planned game times, which was decreased to 6,796 when the confirmed game time was detected on the video. Among the 6,796 events that occurred during the verified game time, 4,021, or close to 60%, were eliminated as they were linked to players who were not on the field during recording and hence not an actual head impact.
This implies that earlier studies, which utilized head impact sensor data without these significant methodological steps, potentially had a high proportion of non-head impact events in their dataset.
The video footage of 1,893 sensor-recorded events for on-the-field players and within the frame of the camera were examined, and the events were divided into three types—non-events in which a player remains stationary (9.6%), trivial events like adjusting a headband (20.9%), and impact events (69.5%).
Head-to-ball contact was the most frequent impact event, representing 78.4% of each impact event. Other impact events included ball-to-head contact (0.8%), player contact (10.9%), and falls (9.8%).
In addition, this study considered both female and male athletes. Female athletes had a higher proportion of trivial events (36.6% versus 14.2%), and a lower proportion of impact events (48.7% versus 78.4%), which may be attributed to constant adjustments of the headband. But among the real impact events, the breakdown of impact types was similar between both genders.
This study enables the field to build on previous research and improve the accuracy of data collected from head impact sensors by emphasizing the importance of detailed confirmation of head impact sensor data. Using head impact sensor data confirmed by video analysis will yield high-quality information from which we can make important health and sports policy decisions to safeguard athletes.
Kristy Arbogast, PhD, Co-Scientific Director and Director of Engineering, Center for Injury Research and Prevention
Arbogast is also the senior author of the study and a professor of pediatrics at the Perelman School of Medicine at the University of Pennsylvania.
With concussion diagnoses in young athletes on the rise, we're grateful for the opportunity to directly participate in this scientific study with Children’s Hospital of Philadelphia. It is our hope that this research could lead to the development of better head sensors for athletes and make athletics safer for future generations.
Katelyn Taylor, Interim Director of Athletics, The Shipley School
The students of The Shipley School collaborated with the CHOP team for this study.
The research was published as part of CHOP’s Objective Translational Multi-domain Early Concussion Assessment Study, financed by the National Institutes of Health (National Institute of Neurological Diseases and Stroke) and the Integrative Science to Advance Pediatric Concussion Diagnosis and Treatment Study, funded by the Pennsylvania Department of Health.