WoW - Wearable Respiration Monitoring Dataset
Description
This dataset was acquired by the Institute of Systems and Robotics, University of Coimbra, as a part of the CMU-Portugal WoW project, Wireless biOmonitoring stickers and smart bed architecture: toWards Untethered Patients (https://inovglintt.com/financiamento/wow/). The main objective of the project is to develop wearable devices for remote and continuous patient monitoring. As a part of this project, we focused on respiration data, to evaluate methods for estimation of human respiration rate through simple and low-cost wearable patches. We are as well interested to understand how body motion, and position (sitting/standing/moving) affects the accuracy of the estimation. Hence, we have recorded data from different sensor inputs. This includes printed, thin-film, and stretchable skin-mounted strain gauges installed on different body locations, bioimpedance measurements via skin-interfacing electrodes, and a chest/belly-mounted accelerometer. This dataset contains respiration data from 4 subjects. In all measurement sessions, we also acquired ground truth data via a nose-mounted thermistor. The dataset can be used along with sensor fusion and machine learning techniques, in order to develop algorithms that can precisely estimate the respiration rate, during different body positions and activities.
Files
Steps to reproduce
The data was acquired to assess the behavior of the developed respiration monitoring patches and its accuracy while acquiring respiration signals. <strong>Patch P1:</strong> Consisted of a strain gauge printed on a polymeric film and connected to PCB with an inertial measurement unit (IMU). Besides the strain gauge data and the IMU, patch P1 also acquired bioimpedance measurements through four electrodes placed on the upper chest of each subject. <strong>Patch P3:</strong> Identical to the first one but is printed on a softer polymeric film, being, therefore, more stretchable. In this acquisition, bioimpedance was not acquired, only the strain gauge measurements and the IMU data. <strong>Patch PD1:</strong> Double strain sensor patch with smaller strain gauges but using the same materials as P3 - softer and more stretchable polymeric film. Depending on the positions and on each subject the respiration can be more thoracic or more abdominal: one strain gauge placed on the ribs to acquire thoracic respiration and other, perpendicular to the first one, on the stomach, to acquire abdominal respiration. In every acquisition of every patch, a thermistor sensor is always used as ground truth. The thermistor changes its resistance with temperature variations: when exhaling the temperature of the air increases and decreases upon exhaling. The data sets are divided in three main folders, one per patch: DataSet_P1, DataSet_PS1, DataSet_PD1. Each DataSet folder is divided in four folders, one per subject. Each subject's folder contains the csv files of each sensor during different tests: <strong>DataSet_P1 Subject Y (Y=1,2,3,4)</strong> TestZ_pos (Z_pos = 1_sitting, 2_standing, 3_fowler): .Bioimpedance data: 'testZ_pos_BioZ.csv' .IMU data: 'testZ_pos_imu_out.csv' .Strain Gauge data: 'testZ_pos_rr_out.csv' .Thermistor data: 'testZ_pos_therm_out.csv' <strong>DataSet_PD1 Subject Y (Y=1,2,3,4)</strong> TestZ_pos (Z_pos = 1_sitting, 2_standing): .IMU data: 'testZ_pos_imu_out.csv' .Strain Gauge data: 'testZ_pos_rr_out.csv' .Thermistor data: 'testZ_pos_therm_out.csv' <strong>DataSet_PS1 Subject Y (Y=1,2,3,4)</strong> TestZ_pos (Z_pos = 1_sitting, 2_standing): .IMU data: 'testZ_pos_imu_out.csv' .Strain Gauge I data: 'testZ_pos_rr_out.csv' .Strain Gauge II data: 'testZ_pos_rr2_out.csv' .Thermistor data: 'testZ_pos_therm_out.csv' <strong>How to read the CSV files: </strong> Each CSV file has a timestamp in the first column, and the values of the sensor in the next column. The IMU files, have 7 columns, comma separated: the first one is the timestamp, and the following ones are the accelerometer data across x, y and z axes, followed by the gyroscope data across x, y and z axes, in this order. You can use any popular software for inspecting and analyzing the data in the CSV files, such as Python libraries (csv.reader/readlines/pandas), Matlab or MS Excel.