Perceptual Decision Efficiency Is Modifiable and Associated with Decreased Musculoskeletal Injury Risk Among Female College Soccer Players

Abstract

Background: Prevention and clinical management of musculoskeletal injuries have historically focused on the assessment and training of modifiable physical factors, but perceptual decision-making has only recently been recognized as a potentially important capability. Immersive virtual reality (VR) systems can measure the speed, accuracy, and consistency of body movements corresponding to stimulus-response instructions for the completion of a forced-choice task. Methods: A cohort of 26 female college soccer players (age 19.5 ± 1.3 years) included 10 players who participated in a baseline assessment, 10 perceptual-response training (PRT) sessions, a post-training assessment that preceded the first soccer practice, and a post-season assessment. The remaining 16 players completed an assessment prior to the team's first pre-season practice session, and a post-season assessment. The assessments and training sessions involved left- or right-directed neck rotation, arm reach, and step-lunge reactions to 40 presentations of different types of horizontally moving visual stimuli. The PRT program included 4 levels of difficulty created by changes in initial stimulus location, addition of distractor stimuli, and increased movement speed, with ≥90% response accuracy used as the criterion for training progression. Perceptual latency (PL) was defined as the time elapsed from stimulus appearance to initiation of neck rotation toward a peripheral virtual target. The speed-accuracy tradeoff was represented by Rate Correct per Second (RCS) of PL, and inconsistency across trials derived from their standard deviation for PL was represented by intra-individual variability (IIV). Perceptual Decision Efficiency (PDE) represented the ratio of RCS to IIV, which provided a single value representing speed, accuracy, and consistency. Statistical procedures included the bivariate correlation between RCS and IIV, dependent t-test comparisons of pre- and post-training metrics, repeated measures analysis of variance for group X session pre- to post-season comparisons, receiver operating characteristic analysis, and Kaplan-Meier time to injury event analysis. Results: Statistically significant (p < 0.05) results were found for pre- to post-training change, and pre-season to post-season group differences, for RCS, IIV, and PDE. An inverse logarithmic relationship was found between RCS and IIV (Spearman's Rho = -0.795). The best discriminator between injured and non-injured statuses was PDE ≤ 21.6 (93% Sensitivity; 42% Specificity; OR = 9.29). Conclusions: The 10-session PRT program produced significant improvement in perceptual decision-making that appears to provide a transfer benefit, as the PDE metric provided good prospective prediction of musculoskeletal injury.

Keywords: cognitive training; injury prevention; perceptual decision-making; performance enhancement; virtual reality.

Figure 1

Flowchart depicting the time course of assessments, training, and injury surveillance.

Figure 2

Depiction of time intervals derived from angular (neck rotation) and translatory (arm reaching) displacements of body segments toward a virtual target located beyond the peripheral field of view of the immersive virtual reality headset display, including definitions of performance metrics. The illustration photo is reproduced with permission from Ref. [59].

Figure 3

Pre- to post-training (i.e., pre-season) and post-season changes in (A) Rate Correct per Second, (B) Intra-Individual Variability, and (C) Perceptual Decision Efficiency for training group (solid lines) and pre- to post-season changes for comparison (no training) group (dashed lines). Numerical values are means derived from original (untransformed) data and error bars define corresponding 95% confidence intervals for the mean values. All pre- to post-training changes and group differences for pre-season and post-season were statistically significant (p < 0.05).

Figure 4

Inverse logarithmic correlation (curved solid line) between the Rate Correct per Second (RCS) and Intra-Individual Variability (IIV) performance metrics for neck perceptual latency at baseline (Spearman’s Rho = −0.795; p < 0.001). White circles identify baseline (pre-training) values for the 10 training group players and Xs identify baseline (pre-season) values for 16 untrained players. Dashed lines identify cut points for RCS (≤1.64) and IIV (≥0.134) that prospectively discriminate players who sustained an injury from those who remained uninjured for the entire season. Black circles identify post-training (pre-season) values for the 10 training group players.

Figure 5

Kaplan–Meier depiction of musculoskeletal injury incidence (first injury event) across a 14-week period. A marginally significant difference (Mantel–Cox Log Rank p = 0.059) was documented between players categorized as low versus high performers on the basis of pre-season Perceptual Decision Efficiency (PDE ≤ 21.6 versus > 21.6).