3D printed leech-inspired origami dry electrodes for electrophysiology sensing robots

The study addresses the challenge of reliable, continuous, cuffless blood pressure (BP) monitoring in remote healthcare settings. While cuff-based sphygmomanometers are unsuitable for continuous monitoring, paired ECG–PPG systems using pulse arrival time (PAT) enable continuous BP estimation. However, ECG sensing requires robust skin contact; wet or adhesive electrodes can cause skin irritation and require replacement, complicating robot-assisted monitoring. Dry electrodes avoid these drawbacks but suffer from poor contact and motion artifacts. The authors propose a leech-inspired origami (LIO) dry electrode that uses an origami structure mimicking the leech’s suction and body expansion to improve conformal skin contact without external belts or adhesives. The goal is to realize reliable ECG acquisition for a humanoid sensing robot and demonstrate cuffless BP monitoring via paired ECG–PPG using PAT.

Prior work to improve ECG electrodes includes conducting polymers (e.g., PEDOT:PSS with Ag/AgCl) to enhance skin interface and microstructured or extra-soft self-adhesive polymers to improve adhesion. Patterned dry electrodes printed on stretchable substrates have also been reported. Despite advances, many approaches still require external supports (belts/tapes) that hinder frequent or comfortable monitoring. Origami structures offer tunable mechanical properties (rigidity, rotation, bending) through folding patterns, with dynamics localized to crease lines, making them attractive for embedding sensors. Bioinspired origami systems have demonstrated suction and foldability for grasping and deployable structures. Leech adhesion involves simultaneous non-auxetic (sucker expansion) and auxetic (body) behaviors, which origami can reproduce in a monolithic structure. These insights motivate an origami-based dry electrode that provides self-adhesive, stable contact for ECG sensing.

Design and optimization of LIO: - Bioinspired architecture: A non-auxetic tip (sucker) combined with an auxetic trunk (body), realized via a three-layer Kresling-type origami to mimic leech suction/expansion behavior. Parametric angles α and β tuned auxetic behavior, rotation, and stiffness. - Parametric study: Evaluated Poisson’s ratio, rotation angle under z-compression (Δh), and elastic modulus through compression tests (Shimadzu EZ-LX). Angular designs compared (α=30°, β=38–41°); β=40° selected as optimal balancing strong rotation/Poisson effects with proper folding stability. Pull-off adhesion force measured. Electrode design and fabrication: - Conductive serpentine dry electrodes printed on the non-auxetic tip plane to accommodate in-plane strains during expansion. Serpentine pattern uses repeating semi-circular lines for stretchability. - Ink: Silver flake-loaded polyurethane binder (PU in DMF/THF) mixed by SpeedMixer; viscosity adjusted by heating. - Printing: Direct writing (Musashi SHOT mini 100Sx) at ~1 mm/s; curing at 80 °C for 1 h. Electrode thickness layered (n=1–3) to modulate t/w and buckling resistance. - Electrical/mechanical characterization: Optical microscopy to track cracking under strain; impedance vs stretch measured on flexible plates. Skin impedance of electrodes with and without LIO measured by LCR meter (20 Hz–400 kHz). Origami fabrication: - FFF 3D printing (Tenlog TI-D3 Pro, direct-drive) of Ninjaflex TPU85A for LIO; nozzle 0.2 mm; speed 30 mm/s; layer height 0.1 mm; temperature 223 °C; width 0.1 mm. Sensor frames printed in co-polyester (Ultimaker 3). Designs prepared in SolidWorks and sliced by Cura. - For pumping/suction demonstration, bottom hole sealed with silicone rubber (Eco-flex 00-30). Robot integration and sensing setup: - ECG: Two thimble-type LIO sensors mounted on the humanoid robot’s (Softbank Pepper) middle fingertips; contact on the subject’s chest near the heart. ECG acquisition via a commercial system (Wellue Pulsebit EX; ±3 mV input range; 0.05–40 Hz bandwidth). 30 s recordings; Bluetooth transfer to robot. Voice-triggered logic initiates sensing and obstacle checking. - PPG: MAX30102 module on the opposite robot hand contacting the subject’s fingertip. ECG and PPG acquired simultaneously for time synchronization. - Subject: Single male (36 years; 178 cm; 88 kg), healthy, consented. Exercises for protocol: ECG—climbing 100 and 200 stairs; BP—push-ups, rest, squats, fast walk, 50 m sprint. BP estimation via PAT: - PAT computed as time between ECG R-peak and onset/systolic characteristic of PPG. Calibration conducted against cuff-based sphygmomanometer readings (brachial, left upper arm) to fit linear models: BP = β0×PAT + β1 for SBP and DBP using least squares. Bland–Altman analysis assessed agreement. Ethics: - Approved by Simon Fraser University Research Ethics (Minimal Risk Approval 2020s0193); written informed consent obtained; confidentiality maintained.

- LIO mechanics and adhesion: - Under compression, non-auxetic tip area increased up to ~1.6× at Δh=8 mm; auxetic trunk area reduced to ~0.4×. Rotation angle increased linearly with Δh, converting axial load to torque for rubbing contact. - Pull-off adhesion force for α=30°, β=40° LIO: 1.21 ± 0.24 N. - Elastic modulus exhibited three regions due to multi-stability; β≥40° increased stiffness; β>41° disrupted folding behavior, guiding choice of β=40°. - Serpentine dry electrodes: - n=3 printed layers maintained conductivity up to strain d/do ≈ 1.8 with minimal path disconnections; n=1 failed near 0.15 strain; n=2 reliable up to ~0.8 strain. Increased t/w raised critical buckling force, improving stretchability. - Skin impedance with LIO ~23% lower than without LIO across frequency range, indicating improved contact. - ECG performance on humanoid robot: - LIO-ECG signal SNR: 21.7 ± 0.56 dB vs 18.1 ± 0.85 dB for wet Ag/AgCl electrodes; comparable waveform amplitudes and clear peaks. - Repeatability over 60 cycles: SNR and ECG amplitude stable within ~20 dB and 0.42 mV variability; after 30-day ambient storage ECG amplitude ~0.41 mV, with longer storage showing reduced output (likely due to dust/contamination). - Exercise responsiveness: Heart rate increased from 86 bpm (rest) to 117 bpm (after 100 stairs) and 122 bpm (after 200 stairs), with corresponding peak timing shifts captured. - Cuffless BP via ECG–PPG PAT: - Calibration yielded linear models (example subject): SBP = −50.23×PAT + 168.12; DBP = −42.04×PAT + 116.42; R²=0.948 (SBP), 0.890 (DBP). - Agreement (Bland–Altman): mean difference (estimated − measured) ≈ −0.03 mmHg (SBP) and 0.00 mmHg (DBP); SD 2.13 mmHg (SBP), 2.58 mmHg (DBP), meeting AAMI criteria (≤5 mmHg mean, ≤8 mmHg SD). - Across conditions (push-ups, squats, fast walk, sprint), estimated SBP/DBP tracked cuff measurements closely (e.g., estimated SBP ~150–167 mmHg vs cuff 149–166 mmHg; DBP ~101–116 mmHg vs cuff 99–115 mmHg).

The LIO integrates leech-inspired suction with an auxetic–non-auxetic origami architecture to enhance dry electrode–skin contact without external adhesives. The structure’s axial-to-torsional force conversion (rotation) and area modulation increase conformal contact and reduce contact impedance, enabling high-SNR ECG comparable to wet electrodes even during post-exercise motion and variable breathing. Robust serpentine electrodes (n=3) accommodate large in-plane strains during LIO expansion, maintaining electrical integrity. With reliable ECG near the heart paired with distal PPG, the system accurately estimates BP via PAT: calibrated linear models exhibited strong correlations and excellent agreement with cuff measurements across diverse exercise states. These findings directly address the need for continuous, cuffless BP monitoring compatible with robot-assisted, remote healthcare, demonstrating that origami-enabled dry electrodes can deliver clinically relevant signal quality with minimal patient burden.

This work introduces a bio-inspired leech-inspired origami (LIO) dry electrode integrated into a humanoid robot for ECG acquisition and cuffless BP monitoring. The origami design (α=30°, β=40°) delivers suction-like adhesion, reduced skin impedance, and robust contact, while serpentine electrodes ensure conductivity under large strains. The robot achieved ECG SNR superior to conventional wet electrodes and enabled accurate BP estimation from ECG–PPG PAT with agreement meeting AAMI criteria in a subject across multiple exercise conditions. The approach supports practical, frequent, and comfortable monitoring for remote healthcare. Future work should expand calibration across larger, diverse cohorts, evaluate long-term stability and biocompatibility, refine materials/geometry for durability and hygiene, and validate clinical performance in real-world settings.

- Single-subject study limits generalizability; broader calibration across diverse populations is needed for universal BP estimation models. - PAT includes pre-ejection period, which may vary with physiological states and confound BP estimation without subject-specific calibration. - Storage-related performance drift: ECG output decreased with longer ambient storage, likely due to dust/contamination on electrodes. - No long-term wear or multi-day clinical validation was performed; robustness to sweat, motion, and environmental factors beyond demonstrated scenarios remains to be assessed.