Tennis serve volume, distribution and accelerometer load during training and tournaments from wearable microtechnology

This study analysed the volume, intensity and distribution of tennis serving between (1) training and tournament mesocycles, and (2) within-mesocycle based on session-type, drill-type and week. Seven elite females wore a GPS device, with in-built accelerometer and gyroscope, during training and tournament mesocycles. Sessions were classified as "practice sessions" or "official matches" and drills as "serve-focused drills" or "point-play". Training weeks were classified according to their timing prior to tournaments ("Week −3", "Week −2" and "Week −1"). Machine learning algorithms detected serves from the raw accelerometer and gyroscope data to report serve count, load and rotation magnitude in absolute and relative (.min−1 or serve) values. Cohen's d with 95% confidence intervals compared serve volume (count), intensity (serve load, rotation magnitude) and distribution (n.min−1) between session- and drill-type during mesocycles. Cumulative serve counts were compared between training weeks. Increased absolute serve load was observed during official matches compared to practice sessions (d = 0.97–1.07[0.69–1.30]). Serve-focused drills in training mesocycles have lower absolute counts and serve load (d = 0.82[0.54–1.34]), though greater relative count and load (d = 1.10–1.87[0.59–2.19]). Trivial-small ES existed between training weeks for cumulative serve count (d = 0.12–0.25[−0.83–1.27]). Session- and drill-level analyses highlight greater mechanical demands in official matches, though lower serve density in point-play.