An Approach to Model Checking the Perceptual Interactions of Medical Alarms

Abstract

The perceptibiliy of auditory medical alarms is critical to patient health and safety. Unfortunately concurrently sounding alarms can interact in ways that can mask one or more of them: render them imperceptible. Masking may only occur in extremely specific and/or rare situations. Thus, experimentation is insufficient for detecting it in all of the potential alarm configurations used in medicine. Therefore, there is a real need for computational methods capable of determining if masking exists in medical alarm configurations. In this work, we present such a method. Using a combination of formal modeling, psychoacoustic modeling, temporal logic specification, and model checking, our method is able to prove whether a configuration of alarms can interact in a way that produces masking. This paper motivates and presents this method, describes its implementation, demonstrates its power with an application, and outlines future developments.

A computationally efficient formal method for discovering simultaneous masking in medical alarms

“…These psychoacoustics will indicate if a single sound can mask another. They were also the basis for previously published results [34,35]. However, the combined masking threshold of multiple concurrent sounds can be greater than the sum of the masking effect of individual maskers.…”

“…However, because increasing the search depth exponentially increased computational time [36], this was not always possible in practice. This version of the method successfully improved upon the older version of the method [34,35] by both being more usable (due to the spreadsheet-based modeling) and more scalable (the computational efficiency of using the optimized lookup tables) [11,36]. In showing this, we analyzed the alarm system evaluated in the early versions of the method [34,35] as well as the alarms from a real telemetry monitoring system, a GE CARESCAPE TM Monitor B850 [31].…”

“…Researchers have used model checking to successfully find and correct human factors issues in automated systems [6,10,20,51,63] and medical systems [3, 4, 7-9, 54, 60]. However, outside of our previous efforts on alarm masking modeling and detection [11,[34][35][36], no work has used model checking to find safety problems associated with human sensation and perception. Below we describe how our previous efforts worked.…”

“…Below we describe how this new approach was realized. After this, we use the new approach to evaluate the alarm configurations reported in previous results [11,[34][35][36] to compare prediction performance and time. We also characterize how the method scales with a series of synthetic test cases.…”

“…To address this problem, we developed a computational method [11,[34][35][36] that can detect masking in configurations of tonal medical alarms. The method uses a novel combination of psychoacoustics and model checking.…”