Stereo-BP : Non-invasive Blood Pressure Sensing With Earables

Leveraging Pulse Time Difference(PTD) Between Left and Right Ear for Non-invasive Blood pressure Sensing in Earables

Frequent blood pressure monitoring with earables has been increasingly explored owing to its importance in diagnosing cardiac health. While previous solutions for blood pressure estimation in earables are uncomfortable and less accurate, the objective of this study is to achieve non-invasive, accurate and cuff-less blood pressure monitoring with PPG sensors integrated into earphones.

To this end, we investigated photoplethysmograph (PPG) signals from the left and right ears and found that blood arrives earlier in the left ear due to the closer distance from the heart. Based on our findings, we propose Stereo-BP — an earable system leveraging the pulse time difference measured between PPG signals from our left and right ear-worn prototypes to estimate blood pressure

Dataset Description

The StrereoBP Dataset contains approximately 8 hours of sensor data collected from 20 participants (approx. 25 mins of data collected for each participant). Data were recorded from 4 sensors on two wearable devices at the chest and ear-helix:


Device Sensors
StereoBP earable prototype Accelerometer
Gyroscope
Photoplethysmography
Zephyr BioHarness 3.0 chest band Electrocardiography

Additionally, StereoBP also contains ground truth blood pressure readings collected at 7 specific time instances for each participant. In addition, Stereo-BP also contains the following participant related information - Age, Gender, Height, Weight, BMI, Chest circumference, Chest-to-head length, chest-to-leftear length, chest-to-rightear length, left-to-right-ear length, upper-body length, Fat percentage, Muscle percentage, Water Percentage and Bone percentage. The ground truth blood pressures were measured using an OMRON cuff-based blood pressure monitor. The weight percentages were measured for each participant using a QardioBase smart weighing scale. For more details, please see the below publication.

Dataset Structure

The dataset is structured as below:


[participant_id]/
┗ <data files>
README.md
Participant_info.csv

Each participant directory contains time synchronized left and right ear PPG signals stored in separate csv files. In addition, each participant directory also contains ECG signals collected from Zephyr BioHarness 3.0 Chest band as a csv file.

If you use our dataset in your work, please cite the original publication below:

Stereo-BP: Non-Invasive Blood Pressure Sensing with Earables
Ananta Narayanan Balaji, Andrea Ferlini, Fahim Kawsar and Alessandro Montanari
In the 24th International Workshop on Mobile Computing Systems and Applications (ACM HotMobile 2023)