The Arousal Meter (AM) is a gauge derived from heart-rate variability designed to measure autonomic arousal. The purpose of this study was to determine the extent to which the AM could differentiate state shifts in arousal that occurred in response to workload changes. A state shift was considered to be a statistically significant change in the level of arousal relative to the level of workload. Participants (n = 56) were engaged in a dual-task paradigm continuously for 31 minutes that consisted of one of two primary tasks - one high workload (shooting game) and one low workload (surveillance task) - paired with a secondary task (mental arithmetic). The experimental paradigm shifted from high workload (shooting) to low workload (surveillance) for time intervals of 30 seconds, 1 minute, 2 minutes, 4 minutes, and 8 minutes. Participants experienced each time interval twice corresponding to each level of workload. Arousal was averaged across each time interval for each workload level. Means between the low and high workload conditions for the 2, 4, and 8 minute intervals were significantly different in the expected direction (t = 2.20, p < .05; t = 3.82, p < .01; t = 5.85, p < .01). These results indicate that the gauge resolution is approximately 2 minutes. Hence, it appears that the AM is able to differentiate tasks from one another if the tasks are greater than 2 minutes in duration. Results are promising considering the type of tasks the gauge is likely to be used with are longer in nature. Possible applications include mitigation of task characteristics to optimize arousal and subsequently performance.
Measuring heart rate variability is an important component of developing human monitoring systems for soldiers of the next century. Unfortunately, even the best sensors are prone to error in active situations. We have developed a system that detects and corrects errors in interbeat interval data in real time. A six to ten second buffer is used to provide context for a set of rules designed to simulate the way a human expert corrects data offline.
Interbeat interval data was gathered from a pool of eighteen subjects with three detection devices used on each subject. Results of the automated correction were compared with human experts to determine the validity of the method. As expected, success varied based on the number of errors in a neighborhood. Isolated errors were corrected with high accuracy, while severely damaged data streams were totally unrecoverable by human or machine. This technique could serve as a crucial component of interbeat interval based monitoring technologies.