# NeuroSwift: computer vision-based system to assess the cognitive–motor speed of soccer players—preliminary findings

## TL;DR

NeuroSwift is a computer vision system that measures soccer players' cognitive-motor speed, showing professionals react faster in some aspects than university athletes.

## Contribution

NeuroSwift introduces an automated, standardized platform for assessing cognitive-motor speed in soccer with ecological validity.

## Key findings

- Professionals had significantly faster visuomotor reaction speed than university athletes.
- University athletes showed faster displacement speed compared to professionals.
- Professionals had higher response capacity, but cognitive-motor speed differences were not statistically significant.

## Abstract

Cognitive–motor speed (CMS) in soccer integrates perceptual–cognitive processing with motor execution, yet many tools lack this integration and have limited ecological validity. NeuroSwift was engineered as a computer vision-based automated analysis platform to standardize tactical stimuli and produce reproducible measurements.

A 3 × 3 interaction surface, front-facing visual stimuli, and HD video were orchestrated by a web application. Twenty-nine players (15 professionals, 14 university athletes) completed 16 scenarios (8 offensive, 8 defensive). Visuomotor reaction speed (VMRS), displacement speed (DS), and response capacity (RC) were obtained, and cognitive–motor speed (CMS = VMRS + DS, in seconds) was computed. Normality and homogeneity were verified using Shapiro–Wilk and Levene’s tests. VMRS and DS were compared using independent-samples t-tests (Bonferroni α = 0.0167). RC and CMS were assessed using the Mann–Whitney U test. Effect sizes were estimated. All tests were two-tailed, and confidence intervals were estimated where applicable.

Professionals showed faster VMRS (0.77 ± 0.12 vs. 0.96 ± 0.12 s; p < 0.001; d = 0.79), whereas university players showed faster DS (0.64 ± 0.06 vs. 0.76 ± 0.11 s; p < 0.001; d = −0.71). RC favored professionals (median 100.00% vs. 93.75%; Z = 3.13; p < 0.001; r = 0.58). CMS tended to favor professionals (median 1.53 s vs. 1.61 s) without significance (Z = −0.544; p > 0.05; r = 0.10).

NeuroSwift enabled standardized stimuli, automated footstep detection, and reproducible in situ laboratory metrics. Expertise was discriminated in perceptual–cognitive and decision components, supporting athlete monitoring, training prescription, and applied research.