FoG-STAR: Multi-Level Annotated Sensor Dataset of Gait Freezing Episodes in Parkinson's Disease (Nature 2026)

Overview

The FoG-STAR (Freezing of Gait - Sensor Technology and Research) dataset published in Nature Scientific Data (January 2026, applicable to 2025 research cycle) provides comprehensive wearable sensor data for detecting and characterizing freezing of gait episodes in people with Parkinson's disease.

Data Collection

• 22 participants with Parkinson's disease experiencing freezing of gait recruited from clinical settings.
• Each participant performed a standardized series of motor tasks known to provoke FoG episodes (turning, narrow passages, dual-task walking, etc.).
• Four inertial measurement units (IMUs) worn simultaneously on both ankles, one wrist, and lower back for comprehensive movement capture.
• Supervised data collection in controlled clinical environment with video recordings for ground truth annotation.

Sensor Measurements

• Tri-axial accelerometer: Linear acceleration in x, y, z axes from all four IMU locations.
• Tri-axial gyroscope: Angular velocity measurements capturing rotational movements.
• Sampling rate: Typically 100-200 Hz for capturing rapid gait transitions and freezing onset.
• Sensor placement: Standardized locations enabling reproducible algorithm development.

Multi-Level Annotations

• FoG episodes: Start and end timestamps of each freezing episode annotated by trained clinicians using video review.
• Severity ratings: Clinical assessment of FoG severity and duration for each episode.
• Motor task labels: Identification of which standardized task (e.g., turning, walking, dual-task) was being performed.
• Pre-FoG states: Annotations of gait changes preceding freezing episodes for early detection research.

Use Cases

• FoG detection algorithms: Developing and benchmarking real-time freezing of gait detection systems for wearable devices.
• Severity assessment: Creating models to quantify FoG severity from sensor data for objective clinical evaluation.
• Prediction research: Identifying pre-freezing gait patterns to enable anticipatory interventions (cueing, warnings).
• Personalized medicine: Analyzing individual FoG patterns to tailor treatment and assistive device strategies.
• Clinical trials: Providing objective outcome measures for evaluating therapeutic interventions (medications, deep brain stimulation, cueing devices).