Annals of biomedical engineering · 2026
Comparison of Six Data Cleaning Methods for Determining Repetitive Head Impact Exposure in Youth Tackle Football.
СтатьяСравнительное исследованиеНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
DeAngelo Samantha, Culiver Adam, Le Flao Enora, Shoaf Nick, Doshi Durshil, Tracy Ryan, Aryeetey Nii-Ayi, Quatrale Anna, Smith Carly, Ma Jianing, Pan Jeff, Yang Jingzhen, Rose Sean C, Onate James, Edwards Nathan, Saygin Zeynep, Caccese Jaclyn B
Аннотация
PURPOSE: Instrumented mouthguards (iMGs) are commonly used to quantify head acceleration event (HAE) exposure, but accurate interpretation requires rigorous data cleaning methods. This study compared six data cleaning methods for determining HAE rates and magnitudes, as well as cleaning method validity compared to the 5th method video verification in youth tackle football.
METHODS: Fifty athletes (ages 8-12) wore Impact Monitoring Mouthguards during games across one season. Six data cleaning methods were applied to HAEs, including uncleaned data, time-windowing, proprietary classification algorithms, video verification, and combinations thereof. Impact rate, peak linear acceleration (PLA), and peak rotational velocity (PRV) were compared across methods using rate ratios, and intra-class correlation coefficients (ICCs), and non-parametric analyses.
RESULTS: Data cleaning methods significantly influenced HAE rate but had minimal effect on magnitude. The uncleaned dataset produced the highest HAE rate (67.75 per athlete exposure), while the most stringent method (i.e., time-windowed, proprietary algorithm-classified, video-verified data) yielded the lowest (0.70 per athlete exposure). Although the time-windowed, proprietary algorithm-classified data demonstrated high specificity (0.96), it demonstrated low sensitivity (0.37) and positive predictive value (0.39) when compared to video-verified data. Differences in PLA across methods were not significant; only one significant difference in PRV was observed.
CONCLUSIONS: These findings highlight the impact of data cleaning on HAE quantification in youth tackle football. Although video verification remains best practice, it is resource intensive. Time-windowed, algorithm-classified data may serve as an efficient proxy in similar cohorts, though researchers should recognize its limitations. Findings support the need for standardized data cleaning methods and transparent reporting to ensure accurate and comparable HAE exposure estimates.
METHODS: Fifty athletes (ages 8-12) wore Impact Monitoring Mouthguards during games across one season. Six data cleaning methods were applied to HAEs, including uncleaned data, time-windowing, proprietary classification algorithms, video verification, and combinations thereof. Impact rate, peak linear acceleration (PLA), and peak rotational velocity (PRV) were compared across methods using rate ratios, and intra-class correlation coefficients (ICCs), and non-parametric analyses.
RESULTS: Data cleaning methods significantly influenced HAE rate but had minimal effect on magnitude. The uncleaned dataset produced the highest HAE rate (67.75 per athlete exposure), while the most stringent method (i.e., time-windowed, proprietary algorithm-classified, video-verified data) yielded the lowest (0.70 per athlete exposure). Although the time-windowed, proprietary algorithm-classified data demonstrated high specificity (0.96), it demonstrated low sensitivity (0.37) and positive predictive value (0.39) when compared to video-verified data. Differences in PLA across methods were not significant; only one significant difference in PRV was observed.
CONCLUSIONS: These findings highlight the impact of data cleaning on HAE quantification in youth tackle football. Although video verification remains best practice, it is resource intensive. Time-windowed, algorithm-classified data may serve as an efficient proxy in similar cohorts, though researchers should recognize its limitations. Findings support the need for standardized data cleaning methods and transparent reporting to ensure accurate and comparable HAE exposure estimates.