Throwing Event Detection using Acceleration Magnitude collected with Wrist-Worn Sensors
Daniel Schweiger, Shelby Critcher, Todd J. Freeborn, Elizabeth E. Hibberd · 2020
Recent advancements to reduce rates of injury and pain related to pitching include the release of “Pitch Smart” guidelines that specify the number of pitches per day and required rest recommendations for youth and high school baseball players. However, monitoring pitch counts for adherence to these guidelines requires significant human resources for tracking individuals throughout training and games. The availability of low-cost tools that can monitor and report pitch count and cumulative loads have the potential to improve adherence to pitch count guidelines, reduce the burden of monitoring on coaching staff, and reduce the chances of injury. In this work, acceleration datasets collected using a wrist-worn sensor during throwing events are explored as a method to identify throwing events. The detection algorithm, using a 18g threshold and 3 second window, identified 100% of the 161 pitching events from 8 participants, with only 1 false-positive. This supports that using acceleration datasets from wrist-worn sensors may be an effective method to identify and count throwing & pitching events.