Voice metrics and workload in helicopter rescue: insights from ninety-six cockpit recordings

Abstract

The relationship between subjective workload and task performance is multifaceted. In aviation research, task performance is typically evaluated using multiple variables. Voice metrics can provide valuable insights regarding operational improvements and offer objective assessments. This study aimed to analyze whether voice metrics can reflect task performance, comparing two different helicopter rescue methods, hoist versus rope operations, in a three-crew Helicopter Emergency Medical Service, focusing on the most demanding phase of the operation, the rescuer insertion. Additionally, it examined the relationship between voice-derived task performance measures and workload, measured with the National Aeronautics and Space Administration Task Load Index (NASA-TLX) and heart rate (HR) measures. Outcome variables included total number of commands and corrections issued, deviations from standardized phraseology, and total duration of the insertion phase. Hoist and rope techniques were employed sequentially in standardized rescue missions, with randomization of rescue method sequence and mission location. Sixty-seven paired insertions (hoist and rope) were eligible for analysis. Linear mixed-models accounted for repeated measures. Among both pilots (n = 4) and physicians (n = 3), voice metrics differed clearly between rescue methods (all p < 0.05): Total commands issued were markedly lower during hoisting (29 [26,32]) versus rope operations (47 [44,50]). Hoisting also reduced mean insertion time by 55 [50,60] seconds. Longer insertion time and more commands and corrections issued were associated with higher NASA-TLX score and (varying between crew roles and rescue methods) HR and HR variability. Findings suggest that linguistic voice-derived task performance measures are feasible, non-invasive indicators of operational demand in helicopter rescue operations.