Scoring Performance on the Y-Balance Test Using a Deep Learning Approach.

Area of Science:

• Biomechanics
• Sports Medicine
• Artificial Intelligence

Background:

• The Y Balance Test (YBT) is a widely used dynamic balance assessment in sports medicine.
• Accurate scoring of the YBT is crucial for evaluating functional movement and injury risk.
• Current YBT scoring methods can be subjective and time-consuming.

Purpose of the Study:

• To develop and evaluate a deep learning approach for automated YBT scoring.
• To estimate the normalized reach distance (NRD) using inertial signals from wearable sensors.
• To compare the performance of the deep learning model against existing methods.

Main Methods:

• A deep neural network, incorporating Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM) layers, was utilized.
• Signal processing techniques were evaluated to extract relevant features from inertial sensor data.
• The model was trained and validated on a dataset of 407 subjects using 10-fold cross-validation.

Main Results:

• The deep learning approach achieved a Mean Absolute Percentage Error (MAPE) of 7.88 ± 0.20% for overall NRD estimation.
• Separate regression models for each YBT direction yielded an average MAPE of 7.33 ± 0.26%.
• The proposed method demonstrated a 10% relative MAPE reduction compared to dynamic time warping and k-NN algorithms.

Conclusions:

• Deep learning offers a robust and accurate method for automated YBT scoring.
• Wearable sensor technology combined with AI can provide objective and efficient balance assessments.
• This approach has the potential to significantly advance the application of the YBT in clinical and sports settings.