Collision detection in rugby union using MEMS and machine learning: Comparison of collision intensity between senior professional and elite U20 players

Abstract

Rugby union is a high intensity contact sport where collision events are key determinants of match outcomes, yet the intensity profiles during the transition from elite under-20 (U20) to Senior professional (SP) levels remain poorly described. This study aimed to (i) characterise collision intensity profiles using wearable microelectromechanical systems across sixteen Elite U20 (U20) and twenty-six Senior professional matches, (ii) compare collision frequency and intensity between SP and U20 players, and (iii) examine positional differences in collision demands within each competition level, employing a Random Forest algorithm to classify collisions into six intensity zones (IZ) and determine Mean Maximal Collision Intensity (MMCI). The results reveal that SP players are exposed to a significantly greater collision frequency than U20 players in nearly all intensity zones (p < 0.05), with the most pronounced differences occurring in high-intensity zones IZ 5 (8.01–10.0 g; +17%) and IZ 6 (> 10.01 g; +36%). Furthermore, SP players experienced a significantly higher MMCI than U20 players (11.3 ± 1.2 g vs. 10.9 ± 1.4 g; p < 0.05). These differences were highly position-dependent, with inside backs and loose forwards showing the largest contrasts in exposure. These findings indicate that the progression to professional rugby is characterized by an increase in mechanical severity per contact event rather than a uniform increase in frequency. Consequently, strength and conditioning coaches should implement progressive, position-specific contact conditioning to safely prepare U20 players for the heightened physical demands of the Senior professional game.