Project Team


Students

Aryan Sarin
Mechanical Engineering
Penn State Berks






Faculty Mentors

Joseph Mahoney
Penn State Berks
Kinesiology and Mechanical Engineering


Allison Altman-Singles
Penn State Berks
Kinesiology and Mechanical Engineering


Truong Xuan Tran
Penn State Harrisburg
Computer Science






Project








Project Video




video player icon




Project Abstract


Running is a prominent activity enjoyed by many with a high impact loading on the feet, which could lead to running-related injuries. To avoid injuries, certain runners change their technique, and changing foot strike is a way to benefit a runner’s performance in the long run. Foot strike is the first form of contact that the foot has when striking the ground. There are three types of foot strike patterns (FSPs): Forefoot, Rearfoot, and Midfoot. These types of patterns show the part of the foot that is having first contact, such as the forefoot being the toes of the foot. Any applications and wearable devices help runners provide inputs such as their heart rate, distance, and pace. Providing precise FSPs could prevent running-related injuries. Groups are beginning to produce studies that use Machine Learning to help detect FSPs. This study develops an algorithm using data from a tibia-mounted uni-axial accelerometer from 58 runners. Using TensorFlow, a Neural Network (NN) and long short-term memory (LSTM) model is created, and a Keras tuner is used to create an optimal model based on the dataset. In every training trial, two test subjects are randomly removed for evaluation with a 101 times max voting loop that consists of randomly splitting training and validation for the model. This unique outlook achieved an average of 86% testing accuracy in the NN model and 66% in the LSTM model. The structure could result in an optimal model that could have a potentially great future application that may help many to avoid injuries.




Evaluate this Project


Use this form link to provide feedback to the presenters, and add your project evaluation for award(s) consideration.