# A Longitudinal Analysis of a Motor Skill Parameter in Junior Triathletes from a Wearable Sensor

## TL;DR

This study tracks how junior triathletes' movement cadence changes over two seasons using wearable sensors and finds that non-linear modeling better predicts changes in swimming and running.

## Contribution

The study introduces a custom peak detection algorithm and demonstrates the effectiveness of non-linear modeling in predicting motor skill changes in triathletes.

## Key findings

- Non-linear modeling outperformed linear modeling in predicting swimming cadence changes.
- Group-level non-linear modeling showed increases in swimming and running cadence across seasons.
- Cycling cadence showed negligible changes regardless of modeling technique.

## Abstract

Purpose: Optimal movement cadence is critical to success in elite triathlons. Therefore, the objective of this research was to investigate group and individual longitudinal changes in movement cadence amongst a group of junior triathletes. Method: Junior triathletes (season 1: n = 4, season 2: n = 11) who were members of the state’s talent development pathway wore a single trunk-mounted inertial measurement unit during triathlon races across two triathlon seasons (October 2021 to April 2023). Sensor data were analysed using both linear and non-linear modelling to identify changes in movement cadence across the three disciplines of the triathlon. This allowed for the differences between the two modelling techniques to be contrasted. A custom automatic peak detection algorithm was used to process and analyse the movement cadence data for each triathlete in each discipline. Results: Non-linear modelling performed significantly better than linear modelling in swimming; however, there were no significant differences in model performance between cycling and running. At a group level, non-linear modelling predicted increases in swimming and running cadence across the seasons. However, negligible changes were observed in cycling cadence across the same period. Conclusions: Meaningful changes in movement cadence can be detected with a single inertial measurement unit and confidently predicted in swimming and running over a competitive season when using non-linear modelling techniques. This approach reflects the non-linear nature of human motor skill development and paves the way for similar applications in other sports.