P Nguyen, C Dunbar, A Guyett, K Nguyen, K Bickley, A Reynolds, M Hughes, H Scott, R Adams, L Lack, P Catcheside, J Cori, M Howard, C Anderson, N Lovato, A Vakulin
{"title":"O072 Simple Vestibular-Occular Motor Assessment as a Predictor of Alertness State and Driving Impairment during Extended Wakefulness","authors":"P Nguyen, C Dunbar, A Guyett, K Nguyen, K Bickley, A Reynolds, M Hughes, H Scott, R Adams, L Lack, P Catcheside, J Cori, M Howard, C Anderson, N Lovato, A Vakulin","doi":"10.1093/sleepadvances/zpad035.072","DOIUrl":null,"url":null,"abstract":"Abstract Introduction Driver fatigue contributes to 2-16% of road crashes, highlighting the need for improved detection of at-risk drivers. We used a novel and brief test of vestibular ocular motor system (VOMS) assessed via virtual reality goggles to predict alertness state and driving simulator performance during 29hr extended wakefulness. Methods 49 individuals (Mean±SD Age, 32.6±12.9, 45% Males) undergone 9hr baseline sleep opportunity followed by ~29hrs extended wakefulness with five 60min driving assessments. Cluster analysis, combining steering deviation and number of crashes were used to split participants into groups of either poor vs good driving performance. VOMS assessment was conducted using virtual reality goggles approximately 10mins before and after each drive. Predictive importance of VOMs metrics were ranked using XGBoost machine learning model, which was then utilized to distinguish between poor vs good driving groups. Model performance was evaluated using a 5-fold cross-validation approach using ROC analysis. Results XGBoost machine learning ranked all 70 VOMS metrics on their importance in predicting driving performance group for each drive. Top 10 metrics from pre-drive VOMS test predicted both daytime driving (tests 1-3, AUC 0.8 [95%CI 0.64-0.93], p<0.001) and night-time driving (tests 4-5, AUC 0.78 [95%CI 0.6-0.95, p<0.001]). Post-driving VOMS assessments also predicted daytime (AUC 0.74 [95%CI 0.53-0.9, p<0.001] and night-time driving (AUC 0.76 [95%CI 0.52-0.94, p<0.001]). Conclusion VOMS assessment show promise as a short and effective assessment of sleepiness to predict driving failure. Future validation in independent samples, sleep disordered population and in-field on-road testing are needed to confirm these promising findings.","PeriodicalId":21861,"journal":{"name":"SLEEP Advances","volume":"30 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"SLEEP Advances","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1093/sleepadvances/zpad035.072","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Abstract Introduction Driver fatigue contributes to 2-16% of road crashes, highlighting the need for improved detection of at-risk drivers. We used a novel and brief test of vestibular ocular motor system (VOMS) assessed via virtual reality goggles to predict alertness state and driving simulator performance during 29hr extended wakefulness. Methods 49 individuals (Mean±SD Age, 32.6±12.9, 45% Males) undergone 9hr baseline sleep opportunity followed by ~29hrs extended wakefulness with five 60min driving assessments. Cluster analysis, combining steering deviation and number of crashes were used to split participants into groups of either poor vs good driving performance. VOMS assessment was conducted using virtual reality goggles approximately 10mins before and after each drive. Predictive importance of VOMs metrics were ranked using XGBoost machine learning model, which was then utilized to distinguish between poor vs good driving groups. Model performance was evaluated using a 5-fold cross-validation approach using ROC analysis. Results XGBoost machine learning ranked all 70 VOMS metrics on their importance in predicting driving performance group for each drive. Top 10 metrics from pre-drive VOMS test predicted both daytime driving (tests 1-3, AUC 0.8 [95%CI 0.64-0.93], p<0.001) and night-time driving (tests 4-5, AUC 0.78 [95%CI 0.6-0.95, p<0.001]). Post-driving VOMS assessments also predicted daytime (AUC 0.74 [95%CI 0.53-0.9, p<0.001] and night-time driving (AUC 0.76 [95%CI 0.52-0.94, p<0.001]). Conclusion VOMS assessment show promise as a short and effective assessment of sleepiness to predict driving failure. Future validation in independent samples, sleep disordered population and in-field on-road testing are needed to confirm these promising findings.