Dataset on the Human Body as a Signal Propagation Medium

Overview: This is a large-scale dataset with impedance and signal loss data recorded on volunteer test subjects using low-voltage alternate current sine-shaped signals. The signal frequencies are from 50 kHz to 20 MHz.

Applications: The intention of this dataset is to allow to investigate the human body as a signal propagation medium, and capture information related to how the properties of the human body (age, sex, composition etc.), the measurement locations, and the signal frequencies impact the signal loss over the human body.

Overview statistics:

• Number of subjects: 30
• Number of transmitter locations: 6
• Number of receiver locations: 6
• Number of measurement frequencies: 19
• Input voltage: 1 V
• Load resistance: 50 ohm and 1 megaohm

Measurement group statistics:

• Height: 174.10 (7.15)
• Weight: 72.85 (16.26)
• BMI: 23.94 (4.70)
• Body fat %: 21.53 (7.55)
• Age group: 29.00 (11.25)
• Male/female ratio: 50%

Included files:

• experiment_protocol_description.docx - protocol used in the experiments
• electrode_placement_schematic.png - schematic of placement locations
• electrode_placement_photo.jpg - visualization on the experiment, on a volunteer subject
• RawData - the full measurement results and experiment info sheets
• all_measurements.csv - the most important results extracted to .csv
• all_measurements_filtered.csv - same, but after z-score filtering
• all_measurements_by_freq.csv - the most important results extracted to .csv, single frequency per row
• all_measurements_by_freq_filtered.csv - same, but after z-score filtering
• summary_of_subjects.csv - key statistics on the subjects from the experiment info sheets
• process_json_files.py - script that creates .csv from the raw data
• filter_results.py - outlier removal based on z-score
• plot_sample_curves.py - visualization of a randomly selected measurement result subset
• plot_measurement_group.py - visualization of the measurement group

CSV file columns:

• subject_id - participant's random unique ID
• experiment_id - measurement session's number for the participant
• height - participant's height, cm
• weight - participant's weight, kg
• BMI - body mass index, computed from the valued above
• body_fat_% - body fat composition, as measured by bioimpedance scales
• age_group - age rounded to 10 years, e.g. 20, 30, 40 etc.
• male - 1 if male, 0 if female
• tx_point - transmitter point number
• rx_point - receiver point number
• distance - distance, in relative units, between the tx and rx points. Not scaled in terms of participant's height and limb lengths!
• tx_point_fat_level - transmitter point location's average fat content metric. Not scaled for each participant individually.
• rx_point_fat_level - receiver point location's average fat content metric. Not scaled for each participant individually.
• total_fat_level - sum of rx and tx fat levels
• bias - constant term to simplify data analytics, always equal to 1.0

CSV file columns, frequency-specific:

• tx_abs_Z_... - transmitter-side impedance, as computed by the `process_json_files.py` script from the voltage drop
• rx_gain_50_f_... - experimentally measured gain on the receiver, in dB, using 50 ohm load impedance
• rx_gain_1M_f_... - experimentally measured gain on the receiver, in dB, using 1 megaohm load impedance

Acknowledgments: The dataset collection was funded by the Latvian Council of Science, project “Body-Coupled Communication for Body Area Networks”, project No. lzp-2020/1-0358.

References: For a more detailed information, see this article:  J. Ormanis, V. Medvedevs, A. Sevcenko, V. Aristovs, V. Abolins, and A. Elsts. Dataset on the Human Body as a Signal Propagation Medium for Body Coupled Communication. Submitted to Elsevier Data in Brief, 2023.

Contact information: info@edi.lv