EarSet: Multi-Modal In-Ear PPG and Motion Data from 30 Participants

A multi-modal dataset containing approximately 17 hours of in-ear photoplethysmography (PPG) and motion data from 30 participants. It includes 3-channel PPG signals, 6-axis inertial measurement unit (IMU) data, and reference vital signs from a medical-grade chest device. The dataset was created by Andrea Ferlini at the University of Cambridge to study the impact of head, face, and body movements on PPG morphology.

Use Cases

• Developing motion artifact filtering techniques based on co-located PPG and IMU data streams.
• Classifying types of head and facial movements based on the 6-axis motion signatures described.
• Analyzing morphological differences in PPG signals across different motion states like walking, running, and facial actions.
• Studying the trade-off between data quality and power consumption for in-ear sensor configurations.

Strengths

• Includes data from 30 participants of mixed gender and ethnicity with a mean age of 28.9 years.
• Provides 18 simultaneous data streams per participant, combining multi-wavelength PPG and 6-axis IMU data.
• Captures 16 specific head and facial movements, plus full-body activities like walking and running.
• Includes ground truth vital signs (heart rate, HRV, breathing rate, ECG) from a medical-grade chest device.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Cambridge

Collection Method

Collected using a custom-built flexible research platform with a novel ear-tip design housing PPG and IMU sensors.