Freezing of Gait Prediction From Accelerometer Data Using a Simple 1D-Convolutional Neural Network -- 8th Place Solution for Kaggle's Parkinson's Freezing of Gait Prediction Competition

TL;DR

This paper presents a simple 1D CNN model trained on accelerometer data to detect freezing of gait in Parkinson's patients, achieving competitive results in a Kaggle competition and highlighting deep learning's potential for real-time FOG detection.

Contribution

Introduces a straightforward 1D CNN approach for FOG detection from accelerometer data, demonstrating competitive performance in a large-scale Kaggle challenge.

Findings

Achieved a mean average precision of 0.356 on the private leaderboard.

Ranked 8th out of 1379 teams in the Kaggle competition.

Shows potential of deep learning for real-time FOG detection.

Abstract

Freezing of Gait (FOG) is a common motor symptom in patients with Parkinson's disease (PD). During episodes of FOG, patients suddenly lose their ability to stride as intended. Patient-worn accelerometers can capture information on the patient's movement during these episodes and machine learning algorithms can potentially classify this data. The combination therefore holds the potential to detect FOG in real-time. In this work I present a simple 1-D convolutional neural network that was trained to detect FOG events in accelerometer data. Model performance was assessed by measuring the success of the model to discriminate normal movement from FOG episodes and resulted in a mean average precision of 0.356 on the private leaderboard on Kaggle. Ultimately, the model ranked 8th out of 1379 teams in the Parkinson's Freezing of Gait Prediction competition. The results underscore the potential of Deep Learning-based solutions in advancing the field of FOG detection, contributing to improved interventions and management strategies for PD patients.