Batch processing of 10,000 h of truck driver EEG data

This paper describes the methods used to acquire and reduce a massive amount of EEG data (Wylie et al., 1990). The description is introduced by a review of a previous effort (Mackie and Miller, 1978). The earlier effort created much of the design philosophy for the second effort. The majority of data in the Paradox database came from 400 trips contributed by 80 commercial drivers driving both day and night revenue-cargo runs of 10 or 13 h each. The sleep and driving EEG data were collected with ambulatory Medilog recorders. Breathing and oxyhemoglobin measures were collected during sleep for sleep-apnea determinations. The sleep and driving-EEG data were placed in raw digitized files (128 samples/second), with the latter also available as compressed-band arrays for 20-s epochs, with associated Rechtschaffen and Kales (1968) manual scoring by polysomnographers for all EEG data. Sleep EEG, subjective driving performance and discrete-task data were also placed in the database, integrated and time-registered to within 1-s accuracy with the driving EEG data. Each truck was extensively instrumented for lateral lane position, steering wheel position, speed, video image of the roadway, and video image of the face. Each driver recorded body temperatures several times per day, provided Stanford Sleepiness Scale readings several times each day, and was connected to the Vagal Tone Monitor while driving. In addition, driving segments were prefaced and followed by the performance of the Critical Tracking Task, the Psychomotor Vigilance Task, and the Code Substitution Task. The database should serve as an international resource from which many investigators may draw data.