Stretchable, hair-compatible, and long-term stable wearable electroencephalography device

Abstract

A stretchable, hair-compatible, wearable electroencephalography (EEG) device and method are disclosed herein for enabling mechanically adaptive, hair‑tolerant, and long‑term stable EEG monitoring with reduced motion sensitivity and setup burden. The device may include a controller configured to receive and output measured EEG signals, wherein the output measured EEG signals are subsequently used for real-time monitoring of a brain activity of the user for a period of time.

Description

Attorney Docket No. 10046-682W018684 WAN STRETCHABLE, HAIR-COMPATIBLE, AND LONG-TERM STABLE WEARABLE ELECTROENCEPHALOGRAPHY DEVICERelated Application

[0001] This application claims priority to, and the benefit of, U. S. Provisional Patent Application No. 63 / 788,962, filed April 15, 2025, entitled “STRETCHABLE-HAIR-COMPATIBLE, AND LONG-TERM STABLE WEARABLE EEG SYSTEM,” the content of which is incorporated by reference herein in its entirety.Background

[0002] Electroencephalography (EEG) is widely used for real-time monitoring of brain activity in both neuroscience research and clinical diagnostics. However, current EEG and related monitoring systems experience several limitations, including poor adaptability to varying head sizes, discomfort during prolonged use, unreliable performance during long¬ term monitoring, and difficulty acquiring high-quality signals from hair-covered regions of the scalp. In particular, current EEG electrodes often fail to achieve consistent electrical contact through hair. Many current monitoring systems, therefore, require shaving the scalp or the use of rigid, constrictive head caps, both of which are impractical for routine or ambulatory use and can cause patient discomfort. Collectively, these limitations reduce the feasibility of EEG for extended monitoring and deployment in real-world environments.

[0003] There is a benefit to developing a wearable EEG device that combines adaptability, comfort, and high signal fidelity for diverse populations.Summary

[0004] An exemplary stretchable, hair-compatible, wearable electroencephalography (EEG) device and method are disclosed for enabling mechanically adaptive, hair-tolerant, and long-term stable EEG monitoring with reduced motion sensitivity and setup burden compared to current EEG technologies. Unlike current wearable EEG systems that suffer from rigid form factors, poor user fit, and unstable signal quality, particularly during long-term use or in hair-covered regions, the exemplary device and method achieve consistent, high-fidelity neural signal acquisition by adapting to head geometry while preserving stable electrode placement.

[0005] Current wearable EEG systems require various headset sizes or rigid adjustment mechanisms and remain prone to uneven electrode contact and user discomfort. The exemplary device and method can mitigate these limitations by employing a mesh-integrated support structure with a Kirigami pattern that provides intrinsic stretchability, facilitating theAttorney Docket No. 10046-682W018684 WAN exemplary device and method to conform to various head morphologies and maintain uniform contact pressure without complex fittings or frequent readjustment,

[0006] Current EEG systems exhibit signal degradation when subjected to motion, mechanical deformation, or repeated wear due to strain-sensitive wiring and rigid electrical connections. The exemplary device and method can maintain stable electrical performance under these conditions by using deformable electrical interconnects integrated into the support structure, thereby reducing motion-induced artifacts and improving robustness during motions.

[0007] The exemplary device and method can also overcome challenges associated with EEG recording in hair-covered regions. Unlike wet electrodes that require preparation and dry electrodes that exhibit high impedance and poor motion tolerance, the exemplary device and method enable low-impedance, stable scalp contact without shaving or discomfort by employing hair-compatible electrode assemblies having a soft, three-dimensional (3D) geometry that navigates hair while distributing contact force to maintain comfort during extended wear.

[0008] Current EEG electrodes exhibit time-dependent performance degradation due to drying, impedance drift, or adhesion loss. The exemplary device and method can maintain consistent electrical performance over extended periods by using ionically conductive electrode materials that retain hydration and maintain interface stability, thereby reducing the need for reapplication or maintenance during long-term use,

[0009] As a result of the above-discussed structural and material configurations, the exemplary device and method can provide signal quality comparable to clinical-grade EEG systems while offering improved adaptability, comfort, and long-term stability. Because of these advantages, the exemplary device and method can be used for EEG monitoring and brain-computer interface (BCI) applications in clinical, research, and everyday environments where current wearable EEG systems are impractical or unreliable.

[0010] In an aspect, a wearable device for measuring electroencephalography (EEG) is disclosed comprising: a stretchable substrate configured to conform to a head region of a user; an interconnect layer including a stretchable electrode array, formed in or on the stretchable substrate, including a plurality of conical electrodes configured to conform to the head region and measure EEG signals therefrom, wherein each of the plurality of conical electrodes includes an ionic hydrogel; and a controller operatively coupled to the stretchable electrode array, wherein the controller is configured to receive and output the measured EEG signals, wherein the output measured EEG signals are subsequently used for real-timeAttorney Docket No. 10046-682W018684 WAN monitoring of a brain activity of the user (e.g., brain-computer interface tasks) for a period of time (e.g., 7, 30 days).

[0011] In some embodiments, the interconnect layer further includes: a set of stretchable interconnects electrically connecting the plurality of conical electrodes, wherein the set of stretchable interconnects is configured to maintain an attenuation value of the measured EEG signals below a predefined threshold attenuation value in response to a deformation (e.g., bending, twisting, stretching) of the stretchable electrode array.

[0012] In some embodiments, the predefined threshold attenuation value is 0.2 decibels (dB).

[0013] In some embodiments, the controller is further configured to: adjust, via an amplifier or a filter, the measured EEG signals; and store, in a data storage, the adjusted measured EEG signals, wherein the stored measured EEG signals are subsequently used for offline or longitudinal analysis of the brain activity.

[0014] In some embodiments, the stretchable substrate is configured with a Kirigami pattern configured to be in a first Kirigami configuration when placed on the head region and in a second Kirigami configuration when stretched from the first Kirigami configuration.

[0015] In some embodiments, the ionic hydrogel of each conical electrode is configured to penetrate hair layers and conformably contact skin of the head region to measure EEG signals therefrom.

[0016] In some embodiments, each of the plurality of conical electrodes defines an outer shell and includes: a top region; a porous central region; and a bottom region (e.g., tip) having a smaller surface area than the top region and including the hydrogel, wherein the bottom region is configured to make contact with a subject’s scalp.

[0017] In some embodiments, each ionic hydrogel is at least partially removable and / or wherein each ionic hydrogel is configured to at least partially move within the outer shell (e.g., in a direction of the bottom region of the conical electrode) when the respective conical electrode is in contact with the subject’s scalp.

[0018] In some embodiments, the stretchable substrate includes a plurality of regions, including a temporal region having a first predefined pattern and a scalp region having a second predefined pattern.

[0019] In some embodiments, the temporal region includes a first plurality of apertures featuring a first horizontal spacing (e.g., 5 millimeters (mm)) and first vertical spacing (e.g., 2.5 mm) in relation to one another, wherein the scalp region includes a second plurality of apertures featuring a second horizontal spacing (e.g., 10 mm) and a second vertical spacingAttorney Docket No. 10046-682W018684 WAN (e.g., 5 mm) in relation to one another that is different from the first horizontal spacing and the first vertical spacing such that the temporal region has a higher aspect ratio than the scalp region (e.g., 4.0: 1.5).

[0020] In some embodiments, the stretchable substrate includes a plurality of apertures to facilitate expansion and deformation of the substrate to adapt to different head sizes.

[0021] In some embodiments, the plurality of apertures includes slashes, lines, circles, ovals and / or geometric shapes.

[0022] In some embodiments, the set of stretchable interconnects is coupled to the plurality of conical electrodes via a plurality of snap connectors.

[0023] In some embodiments, each of the plurality of snap connectors includes a male snap connector attached to the respective conical electrode (e.g., in contact with the respective ionic hydrogel) and a corresponding female snap connector attached to a channel location of the interconnect layer.

[0024] In some embodiments, (i) the ionic hydrogel is formed from glycerol, potassium chloride (KCl), and 2-acrylamido-2-methyl-1-propanesulfonic acid sodium salt (AMPSNa), (ii) each stretchable interconnect includes a liquid eutectic gallium-indium material (EGain), and (iii) the stretchable substrate includes at least a silicone elastomer (e.g., Ecoflex 00-30).

[0025] In some embodiments, the wearable device is used for monitoring and / or treatment of at least one of: Parkinson’s disease, epilepsy, Alzheimer’s disease, stroke, traumatic brain injury, psychiatric disorders (e.g., depression, anxiety, obsessive-compulsive disorder), pain, (e.g., brain stimulation, spinal cord stimulation, or peripheral stimulation), peripheral chronic joint pain, overactive bladder syndrome, sleep disorders (e.g., sleep apnea, redness leg syndrome), carpal tunnel syndrome, visual prosthetics and / or blindness, mood disorders, or cognitive / neurodevelopmental disorders (e.g., research and diagnosis of Autism Spectrum Disorder (ASD), Attention-Deficit / Hyperactivity Disorder (ADHD), Dyslexia, Learning Disorders, Motor Disorders).

[0026] In another aspect, a method of fabricating a wearable device for measuring electroencephalography (EEG) is disclosed comprising: providing a stretchable substrate in a substrate mold; embedding a Kirigami mesh template into the substrate mold to define a structural layout within the provided stretchable substrate; removing the embedded Kirigami mesh template from the substrate mold in response to the provided stretchable substrate being cured; placing a stretchable electrode array including a plurality of conical electrodes on or within the provided stretchable substrate corresponding to the structural layout; and depositing a set of stretchable interconnects into the provided stretchable substrate (e.g.,Attorney Docket No. 10046-682W018684 WAN using a stencil and transfer tape) to electrically connect the plurality of conical electrodes of the stretchable electrode array.

[0027] In some embodiments, the set of stretchable interconnects forms an interconnect layer within the provided stretchable substrate to define conductive pathways between the plurality of conical electrodes.

[0028] In some embodiments, the method described herein further includes: curing a hydrogel within each of the plurality of conical electrodes prior to placing the plurality of conical electrodes on or within the provided stretchable substrate.

[0029] In some embodiments, the method described herein further includes: encapsulating the wearable device with a protective layer (e.g., Ecoflex).Brief Description of the Drawings

[0030] Figs. 1A, IB, and 1C each show an example wearable device for real-time monitoring of brain activity in neuroscience research and clinical diagnostics, in accordance with an illustrative embodiment.

[0031] Fig. 2 shows an example method of fabricating the exemplary wearable device, in accordance with an illustrative embodiment.

[0032] Fig. 3A shows an example implementation of the exemplary wearable device, in accordance with an illustrative embodiment.

[0033] Fig. 3B shows an example fabrication process of the exemplary device, in accordance with an illustrative embodiment.

[0034] Fig. 3C shows an example pattern in the exemplary device, in accordance with an illustrative embodiment.

[0035] Fig. 4A shows the configuration, mechanical characterization, and electrical performance of an experimental mesh-integrated, stretchable, hair-compatible, and wearable device configured for real-time monitoring of brain activity, as described in Figs. 1 - 3.

[0036] Fig. 4B show's the signal fidelity of EGain interconnects of the experimental device under twisting deformation.

[0037] Fig. 4C shows the signal fidelity of the EGain interconnects of the experimental device under bending deformation

[0038] Fig. 4D shows the characterization and evaluation of GKAmp-based elastic electrodes in the experimental device.

[0039] Fig. 4E shows adhesion force-displacement curves of the GKAmp hydrogel after eight cycles of attachment / detachment testing on the skin surface.Attorney Docket No. 10046-682W018684 WAN

[0040] Fig. 4F shows photos of GKAmp hydrogel adhesion on fingers and arms.

[0041] Fig. 4G shows the compressive stress of the GKAmp hydrogel with different crosslinker contents.

[0042] Fig. 4H shows the impedance values of GKAmp hydrogel tested at various KCl concentrations and corresponding frequencies.

[0043] Fig. 41 shows the evolution of impedance frequency response and phase characteristics of the GKAmp hydrogel over 1 to 4 weeks.

[0044] Fig. 4 J shows cell viability of the GKAmp hydrogel with various KCl concentrations after 48 hours of co-culture.

[0045] Fig. 4K shows the EEG signal fidelity of the GKAmp hydrogel and a commercial EEG gel.

[0046] Fig. 4L shows the evaluation of the motor imagery (MI) electroencephalogram (EEG) recording and the signal quality of the experimental device over a month.

[0047] Fig. 4M shows topographical maps of mu-band (e.g., 8-13 Hz) event-related desynchronization (ERD) during motor-imagery (MI) and rest tasks across 5 evaluation sessions of the experimental device.

[0048] Fig. 4N shows topographical maps of beta-band (e.g., 18-30 Hz) ERD during MI and rest tasks across 5 evaluation sessions of the experimental device.

[0049] Fig. 40 shows the anthropometric measurements of the subjects / participants during the evaluation sessions.

[0050] Fig. 4P shows an online closed-loop brain-computer interface (BCI) experiment and the performance metrics of the experimental device over time across sessions of the experiment.

[0051] Fig. 4Q shows the subject-wise sample-wise classification accuracy across the three online evaluation sessions of the BCI experiment.

[0052] Fig. 4R shows the subject-wise BCI hit accuracy across the three online evaluation sessions of the BCI experiment.

[0053] Fig. 4S shows the subject-wise scalp-electrode impedance values across 3 online evaluation sessions of the BCI experiment.Detailed Description

[0054] Some references, which may include various patents, patent applications, and publications, are cited in a reference list and discussed in the disclosure provided herein. TheAttorney Docket No. 10046-682W018684 WAN citation and / or discussion of such references is provided merely to clarify the description of the disclosed technology and is not an admission that any such reference is “prior art” to any aspects of the disclosed technology described herein. All references cited and discussed in this specification are incorporated herein by reference in their entirety and to the same extent as if each reference were individually incorporated by reference.

[0055] Example System

[0056] Figs. 1A, IB, and 1C each show an example wearable device 102 (shown as 102’) for real-time monitoring of brain activity in neuroscience research and clinical diagnostics, in accordance with an illustrative embodiment. As shown, the wearable device 102 includes (i) a Kirigami-patterned stretchable substrate 104 having a stretchable electrode array 108 and a set of stretchable interconnects therein and (ii) a controller 106 operatively coupled to the stretchable electrode array 108.

[0057] Kirigami-Patterned Stretchable Substrate (104). In Figs. 1 A - 1C, the Kirigami-patterned stretchable substrate 104 (shown as 104’) is configured to conform to ahead region of a user. In some embodiments, the stretchable substrate 104 is configured to be in a first Kirigami configuration when placed on the head region and in a second Kirigami configuration when being stretched, twisted, or bent from the first Kirigami configuration. In some embodiments, the stretchable substrate 104 is formed of at least a silicone elastomer (e.g., Ecoflex 00-30).

[0058] The stretchable substrate 104 can include a plurality of regions, including (i) a temporal region having a first predefined pattern and (ii) a scalp region having a second predefined pattern. In some embodiments, the temporal region of the stretchable substrate 104 includes a first plurality of apertures with a first horizontal spacing (e.g., 5 millimeters (mm)) and first vertical spacing (e.g., 2,5 mm) in relation to one another. In some embodiments, the scalp region of the stretchable substrate 104 includes a second plurality of apertures with a second horizontal spacing (e.g., 10 mm) and a second vertical spacing (e.g., 5 mm) relative to one another, which differ from the first horizontal spacing and the first vertical spacing. The temporal region should have a higher aspect ratio than the scalp region (e.g., 4.0: 1.5).

[0059] The plurality of apertures, including the first and second plurality of apertures, can facilitate expansion and deformation of the stretchable substrate 104 to accommodate different head sizes. In some embodiments, each of the first and second plurality of apertures includes slashes, lines, circles, ovals, and / or geometric shapes.Attorney Docket No. 10046-682W018684 WAN

[0060] Stretchable Electrode Array (108). In Figs. 1A - 1C, the stretchable electrode array 108, located on or in the stretchable substrate 104, can include or comprise a plurality of conical electrodes configured to conform to the head region and measure electroencephalography (EEG) signals from the head. In some embodiments, the stretchable electrode array 108 forms an interconnect layer of the stretchable substrate 104. The interconnect layer can comprise an electrode array comprising a plurality of conical electrodes configured to conform to the head region and measure EEG signals.

[0061] Each of the plurality of conical electrodes can include a hydrogel (e.g., an ionic hydrogel) configured to penetrate hair layers and conformably contact the skin of the head region to measure EEG signals 114 therefrom. In some embodiments, each conical electrode defines an outer shell and includes (i) a top region, (ii) a porous central region, and (iii) a bottom region (e.g., tip), with a smaller surface area than the top region, containing the hydrogel (see Fig. 3A, subpanel (e)). In some embodiments, the bottom region of each conical electrode is configured to make contact with the user's scalp. This disclosure contemplates that the hydrogel is not limited to ionic hydrogels and can include other types of hydrogels including, for example, without limitation, non-ionic hydrogels, polyethylene glycol (PEG) hydrogels, polyacrylamide (PAAm) hydrogels, gelatin methacrylate (GelMA), cross-linked hydrogels, organogels, elastomers, aerogels, and / or the like. In some embodiments, the hydrogel can include a hydrogel described in WO 2025 / 116209, incorporated herein by reference in its entirety.

[0062] The hydrogel (e.g., ionic hydrogel) of each conical electrode can be (i) at least partially removable and / or (ii) at least partially move within the outer shell (e.g., in a direction of the bottom region of the conical electrode) when the respective conical electrode is in contact with the user’s scalp. In some embodiments, an ionic hydrogel of each conical electrode includes glycerol, potassium chloride (KCl), and2-acrylamido-2-methyl-1-propanesulfonic acid sodium salt (AMPSNa).

[0063] Stretchable Interconnects (110). In Figs. 1A - 1C, the set of stretchable interconnects 110, located on or in the stretchable substrate 104, is configured to electrically connect the plurality of conical electrodes. In some embodiments, the set of stretchable interconnects is configured to maintain an attenuation value of the measured EEG signals 114 below a predefined threshold attenuation value in response to deformation (e.g., bending, twisting, stretching) of the stretchable electrode array 108. The predefined threshold attenuation value can be below 0.2 decibels (dB), at 0.2 dB, or above 0.2 dB.Attorney Docket No. 10046-682W018684 WAN

[0064] The set of stretchable interconnects 110 can be coupled to the plurality of conical electrodes 108 via a plurality of snap connectors. In some embodiments, each of the plurality of snap connectors includes a male snap connector attached to the respective conical electrode (e.g., in contact with the respective ionic hydrogel) and a corresponding female snap connector (e.g., 316) attached to a channel location of the interconnect layer. In some embodiments, each stretchable interconnect includes a liquid eutectic gallium-indium material (EGain).

[0065] Controller (106). In Fig. 1A, the controller 106, implemented within the wearable device 102, is configured to receive the measured EEG signals 114. The controller 106 is then configured to output the measured EEG signals 114 to an analysis / diagnosis system 112 that is configured to process or visualize the measured EEG signals for real-time monitoring of the brain activity of the user (e.g., brain-computer interface (BCI) tasks) for a period of time (e.g., 1 - 30 days). In some embodiments, the analysis / diagnosis system 112 is implemented in a clinical computing system 118 or a mobile device (e.g., smartphone).

[0066] The controller 106 can be further configured to (i) adjust, via an amplifier or a filter, the measured EEG signals 114 and (ii) store, in an analysis database 116, the adjusted measured EEG signals for subsequent offline or longitudinal analysis of the brain activity.

[0067] In Fig. IB, the controller 106 is implemented within the analysis / diagnosis system 112, which is configured to communicate, via a cable connection, with the stretchable electrode array 108.

[0068] In Fig. 1C, the controller 106 is implemented within a remote computing device 122 located in a cloud infrastructure (e.g., cloud server) having a network interface configured to communicate, via a network 120, with the stretchable electrode array 108.

[0069] Example Usage. In some embodiments, the wearable device 102 is used for monitoring and / or treatment of at least one of Parkinson’s disease, epilepsy, Alzheimer’s disease, stroke, traumatic brain injury, psychiatric disorders (e.g., depression, anxiety, obsessive-compulsive disorder), pain, (e.g., brain stimulation, spinal cord stimulation, or peripheral stimulation), peripheral chronic joint pain, overactive bladder syndrome, sleep disorders (e.g., sleep apnea, redness leg syndrome), carpal tunnel syndrome, visual prosthetics and / or blindness, mood disorders, or cognitive / neurodevelopmental disorders (e.g., research and diagnosis of Autism Spectrum Disorder (ASD), Attention- Deficit / Hyperactivity Disorder (ADHD), Dyslexia, Learning Disorders, Motor Disorders).

[0070] Example MethodAttorney Docket No. 10046-682W018684 WAN

[0071] Fig. 2 shows an example method 200 for fabricating the exemplary wearable device 102, in accordance with an illustrative embodiment. The method 200 is further shown in Fig. 3B.

[0072] In Fig. 2, at step 202, the method 200 includes providing a stretchable substrate (e.g., 104) in a substrate mold. At step 204, the method 200 includes embedding a Kirigami mesh template into the substrate mold to define a structural layout within the provided stretchable substrate (e.g., 104). At step 206, the method 200 includes removing the embedded Kirigami mesh template from the substrate mold when the provided stretchable substrate (e.g., 104) is cured. At step 208, the method 200 includes placing (e.g., attaching, integrating) a stretchable electrode array (e.g., 108) having a plurality of conical electrodes on or within the provided stretchable substrate (e.g., 104) corresponding to the structural layout. At step 210, the method 200 includes depositing a set of stretchable interconnects (e.g., 110) into the provided stretchable substrate (e.g., using a stencil and transfer tape) to electrically connect the plurality of conical electrodes of the stretchable electrode array 108.

[0073] In some embodiments, the method 200 includes curing an ionic hydrogel within each of the plurality of conical electrodes before placing the plurality of conical electrodes on or within the provided stretchable substrate (e.g., 104).

[0074] In some embodiments, the method 200 includes encapsulating the fabricated wearable device with a protective layer (e.g., Ecoflex).

[0075] In some embodiments, the set of stretchable interconnects (e.g., 110) forms an interconnect layer within the provided stretchable substrate (e.g., 104) to define conductive pathways between the plurality of conical electrodes.

[0076] Example Implementation

[0077] Fig. 3A shows an example implementation of the exemplary wearable device 102, in accordance with an illustrative embodiment. In Fig. 3A, subpanel (a) shows a mesh- integrated configuration of the exemplary device 102 for adaptability to diverse head sizes. Subpanel (b) shows an example placement of the exemplary device 102 on a user’s head. Subpanel (c) shows example components of the exemplary device 102, including an encapsulation layer 302, stretchable interconnects 110, an interface connector 304, a substrate 104, and electrodes 108. Subpanel (d) shows an example mesh substrate 104 with Gallium¬ indium eutectic (EGain) interconnects 110 in natural and stretched states. Subpanel (e) shows an example porous, conical, soft, three-dimensional (3D) electrode configuration integrated with GKAmp hydrogel for penetration through hair layers and stable scalp contact. Subpanel (f) shows an example interaction of the electrode 108 with hair-covered scalp regions.Attorney Docket No. 10046-682W018684 WAN Subpanel (g) shows an example long-term neural signal acquisition process using the exemplary device 102 to achieve a closed-loop brain-computer interface (BCI).

[0078] In Fig. 3A, the exemplary device 102 is configured with the Kirigami mesh substrate 104 and Gallium-Indium eutectic (EGaIn) stretchable interconnects 110 (see Fig.3A, subpanel (d)), resulting in good stretchability and mechanical compliance. The Kirigami mesh substrate 104 can enable the exemplary device 102 to accommodate diverse head sizes while maintaining stable electrode contact and signal fidelity. The exemplary device 102 can ensure adaptability across diverse conditions and enhance usability in dynamic applications, such as long-term, repeated electroencephalogram (EEG) recordings (e.g., 114) or EEG¬ based BCI operations. To mitigate hair-induced signal degradation, the exemplary device 102 includes a porous, 3D conical electrode structure integrated with an innovative hydrogel comprising Glycerol, potassium chloride (KCl), and 2-Acrylamido-2-methyl- 1 -propanesulfonic acid sodium salt (AMPSNa), collectively referred to as GKAmp (see Fig. 3A, subpanel (e)). This electrode structure can penetrate hair layers (see Fig. 3A, subpanel (f)) and establish direct contact with the scalp while maintaining low contact pressure for comfortable wear and low electrical impedance for extended recordings. Unlike current hydrogel electrodes, which struggle in hair-covered regions or dry environments, the conical, soft 3D electrodes 108 in the exemplary device can provide mechanical compliance that adapts to different scalp morphologies while exerting sufficient contact pressure for stable signal acquisition.

[0079] The exemplary device 102 can be used for both offline and online longitudinal EEG-based BCI tasks that involve neural signal processing and real-time feedback mechanisms, which demonstrates the practicality and longevity of the exemplary device and its potential for facilitating advanced EEG-based applications, such as prolonged EEG monitoring and adaptive brain-machine interaction (see Fig. 3 A, subpanel (g)).

[0080] Example Device Fabrication. Fig. 3B shows an example fabrication process 200 of the exemplary device 102, in accordance with an illustrative embodiment. In Fig. 3B, subpanel (a) shows a substrate preparation using a mold 310 (e.g., 3D-printed) and Ecoflex 302 (e.g., Ecoflex 00-30) poured onto a polyester (PVC) backing film 312. Subpanel (b) shows an integration of a Kirigami mesh template 314 and placement of female snap connectors 316 aligned with the 10-10 EEG montage channel locations. Subpanel (c) shows the cured substrate 104 after removal of the Kirigami template 314, flipped to expose the back side for further processing. Subpanel (d) shows an application of EGain interconnects 110 using a patterned metal stencil 318 and transfer tape 320 to define the conductiveAttorney Docket No. 10046-682W018684 WAN pathways. Subpanel (e) shows the alignment and attachment of the EGain interconnect 110 to the copper interface connectors 316 on a polyimide film and the substrate 104. Subpanel (f) shows a close-up view of precisely aligned EGain interconnects 110 with copper connectors for reliable electrical interfacing. Subpanel (g) shows an encapsulation of the exemplary device 102 with an additional layer of Ecoflex 302 using a peripheral mold for mechanical stability. Subpanel (h) shows an assembly with GKAmp hydrogel-integrated conical porous electrodes 108 snapped onto the substrate 104, completing the exemplary device 102.

[0081] The fabrication of the exemplary device 102 can begin with the preparation of a polyester film 312 (e.g., McMaster-Carr, USA) as a backing layer (see Fig. 3B, subpanel (a)). A substrate mold 310 with a thickness of 4 mm, matching the dimensions of the female snap connectors 316, can be placed on the backing layer, and Ecoflex 302 (Smooth-On, USA) can be poured into the mold 310. A Kirigami mesh template 314 can then be inserted into the substrate mold 310 to define the structural layout. Female metal snap connectors 316 (e.g., Goochain Technology, China) can be positioned within the openings of the Kirigami template 314, corresponding to the targeted channel locations in the 10-10 international EEG montage (see Fig. 3B, subpanel (b)). When the Ecoflex 302 cures, the Kirigami template 314 can be removed, and the exemplary device 102 can be flipped to expose its back (see Fig. 3B, subpanel (c)). A polyimide- based laser-cut copper interface connector can be attached to the substrate 104 to enable electrical interfacing.

[0082] The EGain interconnects 110 can then be fabricated by stirring EGain liquid metal 319 (e.g., Sigma-Aldrich, USA) with a magnetic stir bar for 30 minutes to create an EGain paste. Using a transfer tape 320 (e.g., USCutter, USA) and a customized patterned metal stencil 318 (e.g., JLCPCB, China), the EGain interconnect layout 110 can be printed onto the transfer tape 320 (see Fig. 3B, subpanel (d)) and subsequently transferred to the substrate 104 (see Fig. 3B, subpanel (e)). Once transferred, the EGain interconnect 110 should be aligned with its corresponding copper connector (see Fig. 3B, subpanel (f)). The copper connectors are configured as interfaces for standard pin connectors and other extension components. To encapsulate the exemplary device 102, a 1-mm-thick peripheral mold (e.g., 3D-printed) can be vertically assembled with the substrate mold 310 and the Kirigami template 314. An additional 1-mm-thick layer of Ecoflex 00-30 (shown as 302) can be poured into the assembly and cured (see Fig. 3B, subpanel (g)). Finally, after releasing the exemplary device 102 from the molds 314, GKAmp hydrogel-integrated conical porous electrodes 108 can be snapped onto the exemplary device 102 to complete its assembly (see Fig. 3B, subpanels (h)).Attorney Docket No. 10046-682W018684 WAN

[0083] Example Kirigami Mesh Substrate. Fig. 3C shows an example Kirigami pattern in the exemplary device 102, in accordance with an illustrative embodiment. In Fig. 3C, subpanel (a) shows an example Kirigami pattern configured for setting on a temporal region, with a horizontal spacing (denoted as LHS) of 5 mm, a vertical spacing (denoted as Lvs) of 2.5 mm, and a cut length (denoted as LCL) of 20 mm, yielding an aspect ratio of 4.0. Subpanel (b) shows an example Kirigami pattern configured for a scalp region, with LHS of 10 mm, Lvs of 5 mm, and LCL of 15 mm, resulting in an aspect ratio of 1.5.

[0084] The Kirigami mesh substrate (e.g., 104) of the exemplary device (e.g., 102) can enable precise tuning of its mechanical properties by adjusting unit cell parameters, including horizontal spacing (LHS), vertical spacing (Lvs), and cut length (LCL). In some embodiments, the Kirigami patterns for the temporal and scalp regions of the exemplary device (e.g., 102) are differentiated to address their distinct functional requirements. For the temporal region, the Kirigami pattern has LCL of 20 mm, LHS of 5 mm, and Lvs of 2.5 mm, yielding an aspect ratio (LCL / LHS) of 4.0. For the scalp region, the Kirigami pattern has LCL of 15 mm, LHS of 10 mm, and Lvs of 5 mm, resulting in an aspect ratio of 1.5.

[0085] Example Synthesis of GKAmp Hydrogel. The GKAmp hydrogel can be synthesized via free radical polymerization. First, sodium 2-acrylamido-2-methylpropane sulfonate (AMPSNa, 4.0 g) and potassium chloride (KCl, 0.1 g) can be dissolved in deionized water under continuous stirring at 500 rpm for 1 min. Ammonium persulfate (APS, 0.06 g) can then be introduced as the initiator, followed by the addition of N, N'-methylenebisacrylamide (MBAA) at varying concentrations (e.g., 50, 100, and 150 μL of 70 mg / mL) as the crosslinker. The solution can be further stirred at 500 rpm for an additional minute to ensure uniform dispersion. Subsequently, 3.2 g of glycerol can be added, and the mixture can be continuously stirred under the same conditions. To enhance homogeneity, the solution can be subjected to ultrasonic homogenization for 10 s using an ultrasonic processor. Prior to hydrogel film preparation, the pre -polymerized solution can be placed in a vacuum desiccator to remove air bubbles and left undisturbed for 5 minutes. The degassed solution can then be transferred into designated molds and polymerized in an oven at 60°C for 4 hours. After polymerization, the hydrogel samples can be demolded and allowed to cool to room temperature before further characterization. A series of GKAmp hydrogels can be synthesized by varying the KCl content (e.g., 0.29, 0.58, and O.88wt%). The resulting hydrogels can be labeled as GKAmpx, where x represents the KCl content relative to the total mass of the hydrogel.

[0086] Experimental Results and Additional ExamplesAttorney Docket No. 10046-682W018684 WAN

[0087] A study was conducted to develop and evaluate an experimental mesh-integrated, stretchable, hair-compatible, and wearable device (e.g., 102) (referred to as “MindStretcH”) configured for real-time monitoring of brain activity for neuroscience research and clinical diagnostics, as described in relation to Figs. 1 - 3.

[0088] Evaluation Results

[0089] Evaluation of Kirigami Substrate and EGain Interconnects. Fig. 4A shows a configuration, mechanical characterization, and electrical performance of the experimental device. In Fig. 4A, subpanel (a) shows a Kirigami mesh substrate, structural layout, and electrode placement of the experimental device. The experimental device integrated the Kirigami template structure with strategically placed cuts to enhance stretchability and mechanical adaptability, controlling deformation and localized expansion to conform to varying head sizes. In Fig. 4A, subpanel s (b) and (c) show close-up views of the patterned substrate of the experimental device before and after tensile deformation, demonstrating how the slits expand into the mesh substrate under strain to enable mechanical elongation without compromising structural integrity.

[0090] Fig. 3C provides detailed views of two Kirigami patterns at different regions of the experimental device. The pattern at the temporal region (see Fig. 3C, subpanel (a)) had horizontal spacing (LHS) of 5 mm, vertical spacing (Lvs) of 2.5 mm, and cut length (Le ) of 20 mm, resulting in an aspect ratio of 4.0. This high aspect ratio ensured greater stretchability to conform to complex contours around the head while maintaining comfort. In contrast, the pattern at the scalp region (see Fig. 3C, subpanel (b)) had LHS of 10 mm, Lvs of 5 mm, and LCL of 15 mm, yielding an aspect ratio of 1.5. This lower aspect ratio ensured that the headset remained securely fitted while causing less deformation than the former one.

[0091] To evaluate the mechanical performance of the experimental device, the study conducted finite element analysis (FEA) simulations under realistic stretching conditions. In Fig. 4A, subpanel (d) shows the simulation results for displacement (bottom left) and stress (bottom right) distributions, with the experimental device stretched under a simulated tensile loading to achieve a controlled expansion matching an ear-to-ear distance of 40 cm. The FEA results confirmed that each EEG channel repositioned to approximate the selected electrode locations in the international 10-10 system, ensuring proper neurophysiological mapping during use. The corresponding stress distribution (see Fig. 4A, subpanel (d), bottom right) revealed stress concentration near the edges of the slits. However, maximum stress values remained below Ecoflex-0030's tensile strength (e.g., ~1.38 MPa), ensuring structural integrity under applied loads. The Kirigami patterns delocalized stress across multipleAttorney Docket No. 10046-682W018684 WAN deformation points. The higher Lvs increased the critical buckling force, while the longer LCL enhanced extensibility. The configuration for the temporal region's higher aspect ratio (LCL / LHS=4.0) facilitated greater elongation than that of the scalp region (LCL / LHS=1.5), which prioritized stability.

[0092] The FEA results aligned with a previous study by Shyu et al. (2015), which demonstrated that kirigami patterns improved the strain capacity of materials by delocalizing stress across multiple deformation points

[0028] . In Fig. 4A, subpanels (e) - (f) compare the experimental device with an EEG cap on a subject with challenging head morphology and hair conditions. The EEG cap failed to conform due to limited stretchability, resulting in poor electrode contact (see Fig. 4A, subpanel (e)). In contrast, the experimental device demonstrated better adaptability, maintaining secure coverage regardless of head size or hair type (see Fig. 4A, subpanel (f)).

[0093] To characterize the electrical performance of the Gallium-Indium eutectic (EGain) liquid metal interconnects within the experimental device, the study conducted a series of tests to evaluate signal consistency and fidelity under varying deformation conditions (e.g., stretching, twisting, and bending)

[0031] . First, the study characterized the electromechanical properties of EGain interconnects. In Fig. 4A, subpanel (g) shows the resistance stability over 500 cycles ofloading and unloading at 100% strain, with an inset magnifying 10 cycles to show the signal quality. The results revealed consistent fluctuations in resistance within a narrow range, underscoring the cyclic durability. In Fig. 4A, subpanel (h) shows the linear resistance-strain relationship during a complete loading-unloading cycle, with a hysteresis of 0.63%, indicating reliable electromechanical coupling and elastic recovery. In Fig. 4A, subpanel (i) shows the current response under randomized strain levels (0%, 24%, 57%’, 73%, and 100%), confirming stable current flow with minimal fluctuations. The performance results demonstrated the mechanical resilience of EGain interconnects, ensuring consistent electrical performance under dynamic strain, a critical feature for EEG systems.

[0094] The study further evaluated the signal fidelity of the experimental device under mechanical deformation, including varying strain conditions (see Fig. 4A, subpanels (j) - (1)). The study set the control signal of interest to have a peak-to-peak amplitude of 20 μV to mimic the typical amplitude of EEG signals. In Fig. 4A, subpanel (j) compares the signal fidelity of three signal conditions (e.g., 5 Hz, 15 Hz, 30 Hz) to control metal wires at various strain percentages (e.g., 0%, 25%, 50%, 75%, and 100%). The overlaid waveforms exhibited consistent signal amplitude and phase alignment, confirming their ability to transmit signalsAttorney Docket No. 10046-682W018684 WAN with high fidelity across deformation states. In Fig. 4A, subpanel (k) quantifies attenuation loss for three EEG frequencies (e.g., 5 Hz, 15 Hz, and 30 Hz), showing minimal attenuation (e.g., <0.2 dB) under all strain conditions. This ensures reliable signal acquisition of weak electrophysiological signals such as EEGs. In Fig. 4A, subpanel (1) shows the signal-to-noise ratio (SNR) under the same conditions, with values exceeding 13, and shows no big difference across strain conditions, except for a 30 Hz signal at 25% strain. This phenomenon could be attributed to the intrinsic characteristics of signal transmission systems, in which higher frequencies were more susceptible to attenuation due to resistive and capacitive effects within the transmission medium

[0032] , This demonstrated the interconnects' robustness in preserving signal clarity even at 100% strain.

[0095] In addition to strain tests, the study evaluated the performance of the EGain interconnects under twisting conditions (e.g., 0-5 twists) to assess their suitability for wearable EEG systems in real-world applications. Fig. 4B shows the signal fidelity of the EGain interconnects under twisting deformation. In Fig. 4B, subpanel (a), the overlaid waveforms at 5 Hz, 15 Hz, and 30 Hz remained consistent in amplitude and phase alignment across most twist counts, indicating stable signal transmission under rotational deformation. In Fig. 4B, subpanel (b), the attenuation loss was below 0.2 dB, emphasizing the robust signal transmission. In Fig. 4B, subpanel (c), the interconnects maintained a consistent SNR above 15 across increasing twist counts, with significant differences observed in a few conditions.

[0096] Similarly, the study evaluated bending deformation over an angle range of 0° to 160°. Fig. 4C shows the signal fidelity of the EGain interconnects under bending deformation. In Fig. 4C, subpanel (b), attenuation remained minimal (e.g., <0.2 dB), even at the most extreme bending angles. In Fig. 4C, subpanel (c) indicates that the SNR remained above 15 across most bending angles, with significant differences observed at 45° and 160° for 5 Hz and 30 Hz, respectively. The evaluation results suggest that the EGain interconnects in the experimental device could transmit electrophysiological signals without loss of fidelity, even under severe mechanical deformation.

[0097] Evaluation of GKAmp Hydrogel and Elastic Electrodes. The study developed an integration of GKAmp hydrogel and 3D-printed conical elastic electrodes for the experimental system, addressing hair interference and prolonged wear stability challenges. Fig. 4D shows the characterization and evaluation of the GKAmp-based elastic electrode in the experimental device. Fig. 4E shows adhesion force-displacement curves of GKAmp hydrogel after eight cycles of attachment / detachment testing on the skin surface. Fig. 4FAttorney Docket No. 10046-682W018684 WAN shows photos of GKAmp hydrogel adhesion on fingers and arms. Fig. 4G shows compressive stress of the GKAmp hydrogel with different crosslinker contents.

[0098] In Fig. 4D, subpanel (a), GKAmp hydrogel includes Glycerol, potassium chloride (KCl), and 2-Acrylamido-2-methyl-l -propanesulfonic acid sodium salt (AMPSNa).GKAmp’s dissociative sulfonate groups form a dense network of mobile charge carriers, ensuring stable ionic conduction. Moreover, the hydroxyl and sulfonate groups form numerous hydrogen bonds with skin amines, contributing to strong adhesion, which is beneficial for maintaining stable contact in regions with dense hair coverage

[0021] ,

[0026] ,

[0034] , Glycerol forms a binary solution with the hydrogel’s water content

[0026] , minimizing evaporation and thereby prolonging both the electrical and mechanical stability of GKAmp over time. Furthermore, KCl regulates ionic strength, maintaining a consistent Debye screening length to optimize charge- transfer dynamics and prevent interface polarization

[0035] ,

[0036] ,

[0099] In Figs. 4D - 4F, GKAmp hydrogel adhered to various skin surfaces, including the fingers and forearm, with a maximum adhesion strength of 1.56 N / cm. Even after eight repeated adhesion cycles, the hydrogel remained stable, demonstrating good reusability. Moreover, GKAmp had tunable mechanical properties, with a tensile strength of 32.95 kPa, an elongation at break of 877%, and an average Young’s modulus of 16.5 kPa, ensuring compliance and conformability at the skin interface (see Fig. 4D, subpanel (c)). When compressed to 30% of its natural thickness, the hydrogel's compressive stress increased from 0.23 N ± 0.10 kPa to 1.19 N ± 0.19 kPa. indicating its ability to accommodate various scalp conditions through controlled crosslinking adjustments (see Fig. 4G).

[0100] The study analyzed the impedance-frequency response of different electrode types to assess their suitability for bioelectrical signal acquisition. Fig. 411 shows the impedance values of GKAmp hydrogel tested at various KCl concentrations and corresponding frequencies. Fig. 41 shows the evolution of impedance frequency response and phase characteristics of GKAmp hydrogel over 1 to 4 weeks. Fig. 4J shows cell viability of the GKAmp hydrogel with various KCl concentrations after 48 hours of co-culture.

[0101] In Fig. 4D, subpanel (d), GKAmp and semi-dry electrodes (shown as “Semi-dry 1” and “Semi-dry2”) showed lower impedance than dry electrodes but higher than wet electrodes, with stable performance across frequencies. The impedance decreased with increasing frequency, consistent with capacitive effects. In Fig. 4D, subpanel (e), and Fig. 4H, the interfacial impedance of GKAmp hydrogel decreased with increasing KCl concentration (e.g., 0.29 wt%, 0.58 wt%, and 0.88 wt%). This trend aligned with the ion-Attorney Docket No. 10046-682W018684 WAN compensation effect of KOI in the hydrogel matrix, where increasing KOI concentration provided more mobile ions to facilitate charge transfer, particularly at low frequencies (e.g., <10 Hz), where ionic conduction dominates. The hydrogel showed minimal degradation in charge-transport properties over 4 weeks (see Fig. 4D, subpanel (f); Fig. 41), suggesting its suitability for prolonged EEG monitoring. The weight retention analysis over 600 hours further revealed that glycerol modulated water loss kinetics, maintaining hydration levels above 80% compared with hydrogels without glycerol, despite fluctuations in environmental humidity (30-55%) (see Fig. 4D, subpanel (g)).

[0102] The study conducted cyclic charge injection capacity (CIC) testing over 1500 cycles. In Fig. 4D, subpanel (h), the GKAmp hydrogel maintained a stable charge injection capacity over 4 weeks, indicating sustained electrochemical performance (see Fig. 4D, subpanel (i)). This stability was attributed to the optimized ionic conductivity enabled by KCl incorporation and the hydrogel’s hydration-retention properties, which together ensured efficient charge transport at the skin-electrode interface. Furthermore, cell cytotoxicity tests showed that GKAmp hydrogel electrodes maintained high cell viability (>75%) across all tested KCl concentrations (0.29 wt%, 0.58 wt%, and 0.88 wt%) after 24h and 48h incubation, confirming good biocompatibility (see Fig. 4D, subpanel (j); Fig. 4J).

[0103] The study conducted eyes-open and eyes-closed tests to compare the accuracy of the electrophysiological signals from GKAmp hydrogel and a commercial EEG gel. Fig. 4K shows the EEG signal fidelity of the GK Amp hydrogel and the commercial EEG gel. In Fig.4K, both hydrogels demonstrated comparable mu band amplitude (see subpanel (a)) and power (see subpanel (b)), with high correlation coefficients for amplitude (e.g., r=0.8062) and power (e.g., r=0.9408). In Fig. 4K, subpanel (c), time-frequency spectrograms showed consistent mu-band patterns between eyes-open and eyes-closed states. The results in Fig. 4K confirmed that GKAmp hydrogel provided equivalent signal fidelity to that of commercial EEG gels.

[0104] To address the limitations of commercial wet EEG electrodes, particularly their restricted applicability to diverse hair conditions and the excessive residual gel left on hair after use, the study developed a flexible, 3D-printed conical support platform for the GKAmp hydrogel. By functioning as an insulating layer for the GKAmp hydrogel, the platform prevented the substantial surface area of the internal load from adhering to hair strands, leaving only the hydrogel tip exposed, creating an electrode suitable for use with hair¬ compatible hydrogel materials. By leveraging the cohesion of GKAmp and the platform, theAttorney Docket No. 10046-682W018684 WAN electrode of the experimental device could navigate through the hair and establish direct contact with the scalp, regardless of hair density or texture (see Fig. 4D, subpanel (k)).

[0105] To enhance wearing comfort, the study developed a porous structure at the tip of the conical mold, which reduced localized scalp stress, thereby improving long-term comfort during extended EEG recordings (see Fig. 4D, subpanels (k) - (n)). Under compression, the porous electrode of the experimental device induced lower scalp surface stress than its non-porous counterpart under identical external force conditions. The porous structure enabled greater deformation of the electrode, increasing the effective contact area and distributing stress more evenly across the scalp surface (see Fig. 4D, subpanel (i)). Additionally, displacement simulations (see Fig. 4D, subpanel (m)) revealed that the porous electrode caused less displacement on the scalp than the non-porous electrode, particularly at the contact periphery, which indicated better stress distribution and reduced pressure points. At both strain levels, the porous electrode exhibited lower stress than the non-porous electrode (see Fig. 4D, subpanel (n)), reducing localized pressure on the scalp. Two-way repeated- measures ANOVA revealed differences between porous and non-porous electrodes across both strain levels (e.g., 10% strain: p = 0.0047, n = 4; 30% strain: p < 0.0001, n = 4) (p denotes probability and n denotes quantity).

[0106] The study tested the porous and non-porous electrodes on a human volunteer’s arm and forehead to evaluate residual redness after 10 minutes of wear. The non-porous electrode left more pronounced redness on the skin than the porous electrode, indicating higher localized pressure. In contrast, the porous electrode exhibited reduced redness, indicating that it distributed pressure more evenly across the contact area. This observation aligned with the finite element simulation and compression test results, further confirming that the porous electrode configuration minimized localized stress and enhanced comfort during prolonged use. By combining the porous, 3D-printed elastic conical electrode with GKAmp hydrogel, the study addressed shortcomings of commercial wet electrodes. The GKAmp-based porous / elastic electrode ensured stable skin-electrode contact, enhanced long¬ term wearing comfort, and prevented gel residue on hair.

[0107] Evaluation of EEG Recording and. Signal Quality. To corroborate the feasibility and robustness of the experimental device for supporting EEG-based brain-computer interface (BCI) over an extended period, the study conducted one-month-long motor-imagery EEG recordings on 6 subjects. Motor-imagery (Ml) is a BCI that elicits reproducible event- related desynchronization (ERD) in sensorimotor rhythms (e.g., mu and beta bands), enabling non-invasive detection of cortical engagement during mental rehearsal of movements

[0037] ,Attorney Docket No. 10046-682W018684 WAN [381. This approach is relevant for longitudinal neurorehabilitation applications, where sustained MI practice can drive neural plasticity and functional recovery in motor-impaired populations

[0012] ,

[0038] .

[0108] Fig. 4L shows the evaluation of the motor imagery (MI) EEG recording and the signal quality of the experimental device over a month. In Fig. 4L, subpanel (a) shows a subject wearing the experimental device to perform a motor-imagery task during the evaluation. In Fig. 4L, subpanels (b) and (c) show the grand average normalized power of the mu-band (e.g., 8—13 Hz) and beta-band (e.g., 15-30 Hz) over time during the MI task. The shaded regions represent the standard error of the mean (SEM), indicating the consistency of trends across subjects. In both frequency bands, the study observed event-related desynchronization (ERD) during the MI task, characterized by a sustained decrease in power. In the first two sessions (first week), the ERDs in the mu band were weaker than in the last three sessions, which could be attributed to the gradual improvement of subjects’ motor-imagery technique

[0039] . Overall, these dynamics indicated consistent engagement of the sensorimotor cortex during MI tasks.

[0109] In Fig. 4L, subpanels (d) - (f) show EEG activity during a resting condition, where participants remained relaxed without engaging in motor imagery. In Fig. 4L, subpanel (d) shows a subject wearing the experimental device while performing the rest task. In Fig.4L, subpanels (e) and (f) show the grand average normalized power of the mu-band and beta¬ band, respectively, during rest. Unlike during MI tasks, no ERD was present in either frequency band during rest. The stability of mu- and beta-band power across time highlighted that resting-state EEG signals remained consistent and unaffected by external tasks, further validating the reliability of the experimental device for capturing baseline neural activity.

[0110] In Fig. 4L, subpanels (g) and (h) show the normalized ERD for the mu- and beta¬ bands at the C3 / C4 electrode sites across the one-month evaluation. The temporal progression of mu-band ERD during motor imagery (MI) tasks revealed changes over the one-month evaluation, underscoring the specificity and reliability of task-related cortical modulation captured by the experimental device. Statistical analysis using a two-way repeated -measures ANOVA revealed effects for the interaction between condition (e.g., MI vs. Rest) and session progression (e.g., p=0.0377), as well as for condition (e.g., MI vs. Rest, p=0.0276) and session progression (e.g., pcO.0001). The analysis highlighted the immediate engagement of sensorimotor cortical areas during motor imagery and the progressive enhancement of cortical modulation over time.Attorney Docket No. 10046-682W018684 WAN

[0111] Post hoc (e.g., Turkey’s test) comparisons demonstrated differences in ERD between Mi-mu and Rest-mu across most sessions. On Day 0 (e.g., p=0.0258), users showed a distinction between the MI and rest conditions, indicating their ability to engage sensorimotor areas from the outset. However, by Day 4 (e.g., p=0.0564, ns), this differentiation became less pronounced, reflecting early variability in task performance. From Day 8 (p=0.0043) onward, the significance and magnitude of mu-band ERD steadily increased, reaching peak differentiation by Day 30 (e.g., p<0.0001). In the first two sessions, the mu-band ERD was shorter and showed a bounded recovery in power during the task phase (see Fig. 4L, subpanel (b)), which was attributed to several factors. Similar to motor learning, first, early sessions may have imposed higher cognitive demands on users, leading to inconsistent engagement of sensorimotor areas and shorter ERD durations. As the evaluation progressed, neuroplastic adaptation facilitated the development of task-specific cortical pathways, enabling sustained ERD during motor imagery. Furthermore, improvements in task familiarity and refinements in mental imagery strategies contributed to the observed trends

[0040] . These changes are corroborated by the topographical maps in Figs.4M and 4N, illustrating a shift from broader, less distinct sensorimotor activation in early sessions to more localized contralateral activation in later sessions. This spatial refinement reflected the progressive development of task-specific neural efficiency. The evaluation results indicated that cortical plasticity mechanisms were involved in learning to control BCls with longitudinal training, a phenomenon in Mi-based BCI studies

[0041] ,

[0042] , The beta-band ERD also exhibited a task-related significant difference, with effects of condition (e.g., motor imagery vs. rest, p=0.0276) and session progression (e.g., p<0.0001). However, the lack of an interaction between condition and session progression (e.g., p=0.5715) indicated that betaband ERD did not demonstrate enhanced sensitivity to longitudinal improvements in motor imagery proficiency, unlike the mu-band ERD, which showed progressive changes over time. Post hoc comparisons indicated strong initial task engagement on Day 0 (e.g., pcO. OOOl) but showed diminished differentiation between Mi-Beta and Rest-Beta in later sessions (e.g., p>0.05). This transient nature of beta-band ERD could reflect its broader role in motor control and inhibition rather than specific engagement during motor imagery

[0043] .

[0112] Fig. 4M shows topographical maps of mu-band (e.g., 8-13 Hz) ERD during MI and rest tasks across 5 evaluation sessions. In Fig. 4M, subpanels (a) - (e) show mu-band ERD during MI tasks over five evaluation sessions, demonstrating progressive refinement of contralateral sensorimotor cortex activation, with stronger and more localized desynchronization in later sessions. In Fig. 4M, subpanels (f) - (j) show mu-band ERDAttorney Docket No. 10046-682W018684 WAN during rest tasks over five evaluation sessions, demonstrating minimal desynchronization and consistent baseline activity, highlighting the specificity of task-related modulation.

[0113] Fig. 4N shows topographical maps of beta-band (e.g., 18-30 Hz) ERD during MI and rest tasks across 5 evaluation sessions. In Fig. 4N, subpanels (a) - (e) show beta-band ERD during MI tasks over five evaluation sessions, demonstrating broader and less localized desynchronization than the mu-band, with moderate task-specific modulation. In Fig. 4N, subpanels (f)–(j) show beta-band ERD during rest tasks across five evaluation sessions, demonstrating diffuse, weak desynchronization, further emphasizing reduced task specificity in the beta-band compared to the mu-band. In Fig. 4N, topographical maps reveal a less focused spatial pattern for beta-band ERD, further suggesting its weaker task specificity [44-46],

[0114] In Fig. 4L, subpanels (i) and (j), the topographical maps provided spatial insights into cortical activity during motor imagery (MI) and rest, complementing the temporal trends in Figs. 4L, subpanels (b), (c), (e), and (f). For the mu-band (see Fig. 4L, subpanel (i); Fig.4M), motor imagery elicited strong contralateral (C3) desynchronization over the sensorimotor cortex, which reflected the specific recruitment of motor-related cortical regions during the mental rehearsal of movement. The minimal ipsilateral (C4) desynchronization in the mu-band was consistent with previous studies, suggesting that mu-band activity was driven by contralateral sensorimotor activity and exhibited only weak bilateral involvement

[0047] . The rest condition displayed negligible desynchronization, further emphasizing the taskspecific modulation of the mu-band during motor imagery. In contrast, in Fig. 4L, subpanel (j), the beta-band (e.g., 18-30 Hz) topographical maps exhibited broader, less spatially distinct patterns of desynchronization. Contralateral desynchronization over the sensorimotor cortex, evident during MI tasks, was less pronounced than in the mu-band, which was consistent with the beta-band’s broader functional role in motor planning, control, and inhibition

[0044] . During rest, the beta-band displayed diffuse, modest desynchronization across both hemispheres, indicating lower task specificity. This aligned with the temporal results in Fig. 4L, subpanel (h), and Fig. 4N, where beta-band ERD was significant during early sessions but showed weaker session-to-session improvement than the mu-band

[0043] . Overall, the experimental device's ability to capture these changes across sessions underscored its suitability for extended BCI applications and its potential for broader applications in neural modulation monitoring.

[0115] To compare the signal quality of the experimental device with an EEG cap, the study analyzed trial rejection rates at various amplitude thresholds

[0048] . The trial rejectionAttorney Docket No. 10046-682W018684 WAN rates, as shown in Fig. 4L, subpanel (k), were comparable between the experimental system and the EEG cap with wet conductive Ag / AgCl electrodes (referred to as a control system) across a range of amplitude thresholds (e.g., 50-150 pV). Statistical analysis using a two-way- repeated -measures ANOVA revealed no interaction between rejection threshold and device type (e.g., p = 0.9979). While there was an effect of device type (e.g., p = 0.0287), Sfdak’s multiple comparisons test did not detect differences at any individual threshold level (e.g., p>0.05), suggesting that the proportion of clean trials relative to all recorded trials of the experimental device was similar to that of the control, In Fig. 4L, subpanel (l) shows the root mean square (RMS) amplitude of clean trials, which reflected the signal quality of the clean trials. The RMS amplitudes for the experimental device were, on average, higher than those of the EEG cap, but the difference was not significant (e.g., p = 0.7922), which indicated that the experimental device captured neural signals without introducing noise or signal distortion, confirming its utility in capturing high-fidelity EEG signals across evaluation conditions.

[0116] The study developed a composite metric to evaluate the signal quality of the experimental device, which integrated trial rejection rate and RMS signal quality into a single measure. In Fig. 4L, subpanel (m) shows composite metric scores for the one-time control measurement and the measurements across five sessions (e.g., Days 1, 4, 8, 15, and 30) using the experimental device. Given the nature of the composite metric, lower values indicated better overall signal quality. For most subjects that performed recordings on both systems, statistical analysis revealed no differences between the experimental device and the one-time control system measurement at most time points (e.g., p > 0.05). However, the study observed differences between the control and the experimental device at independent time points in subject 3, demonstrating that while the experimental device maintained robust signal quality comparable to the EEG cap at most time points, there were session-specific variations in performance. The session-specific variations could reflect factors such as variability in electrode-skin contact quality, operator consistency during device application, or subject-specific conditions at certain time points. The session-specific variations showed no consistent trend across subjects or sessions, indicating that the experimental device delivered reliable signal quality over extended recording periods.

[0117] In Fig. 4L, subpanel (n) shows the scalp-electrode impedance values for the experimental device across five recording sessions (e.g., 1, 4, 8, 15, and 30 days). Each data point represented the impedance of one of the 13 channels, accumulated across all subjects. The results demonstrated a gradual reduction in impedance over time, with lower impedance values in later sessions than in the initial ones. Although the two-way repeated measuresAttorney Docket No. 10046-682W018684 WAN ANOVA indicated no interaction effects (e.g., dayl vs day4: p= 0.0632, day 4 vs day 8: p=0.9361, day 8 vs day 15: p=0.9821, day 15 vs day 30: p=0.7618), post hoc analysis (e.g., Tukey's test) identified differences between session 1 and later sessions (e.g., day 8, p=0.0060; day 15, p=0.0008; day 30, pcO. OOOl). This could be because, as the evaluation progressed, the operator improved their technique with the experimental device, achieving more consistent alignment and pressure distribution across all 13 channels, leading to better electrode-skin contact and lower impedance values. Additionally, the hydrogel electrodes were rinsed with water between sessions to remove residual dirt and gel, which could soften the electrode tips, allowing them to better conform to the scalp’s surface during subsequent sessions. Enhanced conformity increased the effective contact area, thereby reducing impedance.

[0118] Fig. 40 shows the anthropometric measurements of the subjects during the evaluation. In Fig. 40, the head circumference and ear-to-ear length of the study participants (e.g., n = 6, female: male = 1:1) had mean values of 57 cm for circumference and 40 cm for ear-to-ear length. The low variance in head size across subjects indicated that both systems were applied under comparable geometric constraints, thereby reducing potential impedance variability caused by differences in head dimensions. While the overall variance in head dimensions across subjects was low, the EEG cap could not be worn by the subject with the largest head circumference. This highlighted a challenge with the EEG caps, which lacked sufficient adaptability to accommodate diverse anthropometric characteristics. In contrast, the experimental device fitted all subjects, demonstrating its adaptability to a wide range of head sizes and hair conditions. The adaptability of the experimental device underscored the advantages of the Kirigami mesh substrate and the use of EGain, ensuring compliance with users under variable geometric constraints and making the device well-suited for diverse populations and real-world applications.

[0119] Performance Evaluation in Closed-Loop Motor Imagery-Based BCI Operation. The study further validated the performance of the experimental device in an online closed- loop BCI system that required continuous, real-time interaction between the user's brain activity and the experimental device. Fig. 4P shows the online BCI experiment and the performance metrics of the experimental device over time across sessions of the experiment.

[0120] In Fig. 4P, subpanels (a) - (f) show the machine learning-driven closed-loop motor imagery-based BCI experiment, including EEG acquisition, preprocessing, covariance matrix computation, Riemannian geometry-based classification, and real-time feedback. In Fig. 4P, subpanels (g) - (i) show the performance metrics of the experimental device in theAttorney Docket No. 10046-682W018684 WAN online BCI experiment. Fig. 4Q shows the subject-wise sample-wise classification accuracy across the three online evaluation sessions of the BCI experiment. Fig. 4R shows the subjectwise BCI hit accuracy across the three online evaluation sessions of the BCI experiment. Fig.4S shows the subject- wise scalp-electrode impedance values across 3 online evaluation sessions of the BCI experiment.

[0121] To evaluate the performance of the online brain-computer interface (BCI), the study used three metrics: sample-wise classification accuracy (see Fig. 4Q), bar dynamics, and BCI hit accuracy (see Fig. 4R). Sample-wise classification accuracy measured the percentage of individual EEG samples correctly classified into their respective classes, providing a direct assessment of the classifier's effectiveness independent of temporal thresholds. In contrast, bar dynamics reflected the user's control over the BCI system during motor imagery tasks by calculating the proportion of a trial's time during which the accumulated evidence supported the correct class, emphasizing sustained user control over time. Meanwhile, BCI hit accuracy evaluated command delivery performance at the trial level by calculating the ratio of correct threshold hits to total threshold crossings, excluding timeout trials. While sample-wise accuracy focused on granular classification performance, bar dynamics captured temporal control dynamics during task execution, and BCI hit accuracy evaluated system-level command reliability across sessions. Together, the 3 metrics provided a comprehensive evaluation framework for understanding classifier precision, user interaction quality, and system-level command delivery in online BCI applications.

[0122] In Fig. 4P, subpanel (g) shows sample-wise classification accuracy across three online evaluation sessions. The repeated-measures ANOVA revealed no differences across sessions (e.g., p = 0.1226), indicating stable classification performance over time. The group- level averages were about 69% for session 1 (e.g., day 1), 68% for session 2 (e.g., day 2), and 73% for session 3 (e.g., day 26 ± 2), indicating that the classifier distinguished between rest and dominant hand Ml classes. Subject-specific results in Fig. 4Q highlighted variability among subjects, with certain subjects demonstrating improved classification accuracy throughout the experiment. For instance, Subject 2 showed improvements in classification accuracy from session 1 to session 3 (e.g., p < 0.001), while Subject 4 showed increases from session 2 to session 3 (e.g., p < 0.01). The subject- specific results aligned with previous studies, indicating that individual learning effects or increased familiarity with the system could contribute to enhanced performance for certain users [9]. In Fig. 4P, subpanel (h) shows the BCI hit accuracy results, reflecting the accuracy of the commands delivery during online runs. Group-level averages were 74%’ in session 1, 71% in session 2, and 77% inAttorney Docket No. 10046-682W018684 WAN session 3. The repeated-measures ANOVA indicated no differences across sessions (e.g., p = 0.2755) over 1 month, highlighting the robustness of the BCI system in translating neural activity into actionable outputs. While the study observed individual variability, most subjects maintained high levels of performance across all sessions.

[0123] In Fig. 4P, subpanel (i) shows the bar dynamics metric results, with group-level averages of 76% for session 1, 74% for session 2, and 78% for session 3. The repeated-measures ANOVA showed no differences between sessions (e.g., p = 0.1957) over 1 month, indicating stable user control over the bar feedback system. Individual trends suggested that some subjects improved their control over time, potentially benefiting from increased familiarity with the task and feedback mechanisms. Fig. 4P, subpanel (j), and Fig. 4S shows impedance measurements across three online evaluation sessions as a measure of electrode¬ skin contact quality and overall signal acquisition stability for the recording sessions. In Fig.4P, subpanel (j) shows a group-level average of 46.3 kΩ across all three sessions (e.g., session 1: 42.2 ± 30.6 k£2, session 2: 47.7 ± 41.5 kl 2, session 3: 48.9 ± 27.1 k£2). The repeated-measures ANOVA revealed no differences between sessions over a month (e.g., p =0.3938).

[0124] The evaluation results demonstrated that the experimental device, together with the applied Riemannian geometry-based Minimum Distance to Mean (MDM) classifier [9],

[0039] ,

[0049] , provided robust and consistent performance over a month at the group level. The stable classification accuracy, BCI hit accuracy, and electrode-skin impedance demonstrated the reliability of the experimental device. The bar dynamics metric further underscored the effectiveness of user control in distinguishing between rest and dominant-hand MI classes in real time. Individual variability in three metrics suggested that user-specific factors played a role in performance outcomes. For some subjects, improvements between sessions could reflect learning effects or increased familiarity with the BCI system. The consistency of group-level results across evaluation sessions also suggested that the experimental device was resilient to temporal non-stationarities in EEG signals. This could be supported by the stability of the experimental device to reduce hardware non-stationarity, and by the Generic Recentering (GR) mechanism, which updated class prototypes after each run to account for variability in neural activity patterns.

[0125] Evaluation Procedure

[0126] To obtain the above-discussed evaluation results and performance metrics of the experimental device, the study conducted a series of simulations (e.g., FEA) and evaluation sessions (e.g., BCI).Attorney Docket No. 10046-682W018684 WAN

[0127] Finite Element Analysis (FEA) Simulation. The study conducted FEA simulations of the experimental device in Autodesk Fusion 360's Simulation mode to evaluate its mechanical performance under tensile loading. The study specified the following material properties for the experimental device: Young’s modulus = 125 kPa, Poisson’s ratio = 0.49, and ultimate tensile strength = 1.38 MPa. The Kirigami template was modeled using parabolic solid elements, resulting in 150,771 nodes and 93,153 elements to ensure computational accuracy. Boundary conditions included fixing both ends of the device and its top and bottom faces, while allowing controlled displacement to simulate an ear-to-ear distance of 40 cm. The study applied a force across multiple regions to match the experimental scenario, with a maximum magnitude of -1.2 N along the x-axis. The outputs analyzed included displacement distributions and stress concentrations.

[0128] FEA simulations of the electrodes were conducted in COMSOL Multiphysics to evaluate the mechanical performance of porous and non-porous elastic 3D-printed electrode molds under compression. The material properties of the electrodes were set to align with the 3D-printed material, with a Young’s modulus of 2 MPa, a Poisson’s ratio of 0.49, and a density of 1010 kg / m3. The scalp was modeled as a soft tissue with a Young's modulus of 300 kPa, a Poisson's ratio of 0.49, and a density of 1100 kg / m3. A linear-elastic material model was applied to both materials.

[0129] The electrode geometries were meshed using free tetrahedral elements, generating 594,726 degrees of freedom to ensure computational accuracy. Boundary conditions included fixing the base of the electrode molds while allowing the tip to deform freely under compression. A uniform downward displacement was applied to simulate scalp interaction under realistic EEG system conditions. The outputs analyzed included von Mises stress distribution and displacement fields on the scalp surface.

[0130] Electromechanical Testing. Electrical response under strain was recorded using the chronoamperometry method via a potentiostat (e.g., SP-300, BioLogic, France). All mechanical tests were performed using a custom tensile testing setup adapted from a reported protocol by Hsieh et al.

[0026] ,

[0050] . A modified 3D-printed holder secured the base of the interconnect samples, while the opposite end was attached to the Arduino-controlled (e.g., UNO REV3, Arduino) motor-driven extension mechanism (e.g., ramp rate of 68 mm / min). The potentiostat was connected to the interconnect via the two-electrode method. The working electrode (WE) was connected to one end of the interconnect. A 10 Q resistor was placed in series with the interconnect to regulate current during testing and connected to the counter / reference (CE / RE) of the potentiostat. The potentiostat was set to an initial appliedAttorney Docket No. 10046-682W018684 WAN voltage of 1 V, and measurements were taken at a sampling frequency of 10 Hz. Strain values were recorded in real time along with the current and electrode voltage.

[0131] Deformation Evaluation. To achieve EEG-mimicking signal levels, the EGain interconnect was incorporated into a grounded, shielded voltage divider circuit that attenuated the input signal to biologically relevant levels. The voltage divider circuit included 5 and 1 Mil resistors, forming a divider that reduced a 4V sinusoidal input signal from a function generator to the microvolt range (e.g., ~20 pV). The interconnect was mounted in a 3D-printed holder, ensuring a stable connection for integration with an EEG amplifier / recorder (e.g., ANT Neuro, Netherlands). The voltage divider circuit incorporated grounding and shielding to minimize electromagnetic and power-line interference, ensuring a fair comparison across temporal conditions within experimental groups.

[0132] Signal measurements were recorded across 5 Hz, 15 Hz, and 30 Hz. The interconnects were subjected to controlled mechanical deformations, including stretching, bending, and twisting, to simulate real-world conditions in wearable EEG applications.During stretching, the interconnect was incrementally elongated to strain levels of 0%, 25%, 50%, 75%, and 100% using 3D-printed modules with predefined lengths, which were added to the 3D-printed holder. During bending, the interconnect was bent to 0°, 45°, 90°, 120°, and 160° using 3D-printed modules to create the desired angles. During twisting, the interconnect was twisted to half, full, double, and five full rotations. Each deformation condition was recorded for 5 seconds at each frequency.

[0133] Signal Fidelity under Deformation Evaluation. Hysteresis (denoted as H) was evaluated for electromechanical coupling and elastic recovery behavior of the EGain interconnects. To acquire hysteresis, a complete strain cycle (100%) was performed and recorded. Hysteresis was calculated using Equation 1.(Eq. 1)

[0134] In Equation 1, E is strain (%), Reading, and Runioading correspond to resistance values during the loading and unloading phases of the strain cycle, respectively.

[0135] The study collected all the signals using LabStreamingLayer

[0051] , The collected data of each deformation condition (format and level) and frequency were segmented, postprocessed, and analyzed. Specific process and analysis steps included notch filtering (60 Hz),Attorney Docket No. 10046-682W018684 WAN normalization, and segmentation for comparison across conditions (within deformation format).

[0136] Attenuation loss was calculated, as shown in Equation 2, for each deformation level as the reduction in signal power from baseline (original signal).(Eq. 2)

[0137] In Equation 2, Ptransmittedrepresents the baseline input power, and Preceivedrepresents the signal power received by the amplifier under each deformation. Equation 2 was repeated for all deformations and frequencies to quantify power attenuation.

[0138] Signal-to-Noise Ratio (SNR) was calculated by comparing signals collected by deformed EGain interconnects to an ideal baseline, as shown in Equation 3, quantifying the impact of deformation on signal integrity. The ideal baseline signal was established by¬ recording the signal generated by the signal generator through a medical-grade EEG lead (Bio-Medical, USA) and feeding it into the amplifier, thereby providing an accurate reference for subsequent measurements.(Eq. 3)

[0139] In Equation 3, PSignalrepresents the recorded signal from the EGain interconnect and Pnoiserepresents the noise power, determined by subtracting the baseline reference signal from the corresponding experimental signals.

[0140] Mechanical Characterization of the GKAmp hydrogel. To evaluate the mechanical properties of GKAmp hydrogels, the study conducted tensile and compression tests using a motorized testing platform fabricated via 3D printing (e.g., i3 MK3S+, Prusa Research) and integrated with a force gauge (e.g., FB200, Torbal). The force gauge had a maximum load capacity of 200 N, and uniaxial strain was applied at a controlled displacement rate of 68 mm / min.

[0141] Prior to testing, the load cell was calibrated to ensure accurate force measurements. The testing platform was controlled by an Arduino-based control unit (e.g., UNO REV3). For tensile testing, the study prepared GKAmp hydrogel samples measuring 20 mm (width) x 35 mm (length) x 2 mm (thickness) to ensure consistency across evaluations. The effective gauge length of each sample was maintained at 15 mm, with 10 mm of each longitudinal end attached to non-stretchable adhesive tape to facilitate clamping onto theAttorney Docket No. 10046-682W018684 WAN force gauge and platform. Nominal strain (ε) was defined as the ratio of length change to the original length of the hydrogel sample. The elastic modulus was determined from the initial linear region (e.g., 10-15%) of the stress-strain curve by calculating the average slope. For compression testing, cylindrical GKAmp hydrogel samples with a diameter of 8 mm and a height of 8 mm were utilized. The test was conducted until a terminal strain of 30% was reached. The nominal stress (σ) was computed as the applied force divided by the original cross-sectional area of the sample.

[0142] To evaluate the adhesion strength of GKAmp on human skin, a 90° peel test was performed using the motorized testing platform. Each GKAmp sample was fabricated with dimensions of 20 mm (width) x 35 mm (length) x 2 mm (thickness). Before testing, the hydrogel was mounted onto a Kapton film backing layer to provide mechanical support. The sample was then pressed onto the forearm of a human subject using a 500 g weight for 10 seconds to ensure uniform contact. Each sample underwent 8 attachment-detachment cycles. The adhesion strength was quantified by dividing the maximum shear force by the corresponding sample width.

[0143] Impedance Measurement. To evaluate the skin-electrode impedance performance of GKAmp hydrogels with varying KCl concentrations, the study compared their impedance with that of commercial dry and wet electrodes using an electrochemical workstation (e.g., SP-300, BioLogic). The study recorded impedance spectra over a frequency range of 1–100 kHz. The study prepared hydrogel samples by casting the prepolymer solution into a 3D-printed rectangular mold (e.g., 25 mm x 25 mm x 3 mm) to ensure uniform sample geometry and volume. For electrical connection, two gold bars (e.g., 10 mm x 20 mm) were embedded at opposite ends of the hydrogel sample, serving as extended probe terminals. The probe terminals were secured using alligator clips for direct connection to the electrochemical workstation. Ten sets of impedance data were recorded per frequency decade, and the corresponding phase-angle curves were obtained when applied to human skin.

[0144] GKAmp hydrogels, dry electrodes (e.g., OPENBCI Shop), Kendall 200-series foam electrodes (Semi-dry 1), and Kendall disposable surface electrodes H124SG (Semi-dry 2) were affixed to the forearm of a subject as the working electrode. A reference electrode and a counter electrode were placed at distances of 10 cm and 20 cm from the working electrode, respectively. All measurements were conducted on the same subject to ensure consistency. The impedance values obtained were normalized by the electrode-skin contact area to determine the specific skin impedance per unit area. To investigate the long-termAttorney Docket No. 10046-682W018684 WAN stability of the hydrogel, impedance measurements were repeated weekly for 4 weeks using the same methodology.

[0145] Environment Stability Evaluation. To evaluate the effect of relative humidity (RH) on hydrogel stability, the study monitored the variation of GKAmp hydrogels with and without glycerol for 600 hours under ambient conditions. The initial mass of each hydrogel sample was recorded after preparation, and subsequent mass measurements were taken at defined time intervals. Mass values were normalized to the initial mass to calculate the relative mass change over time. Concurrently, ambient humidity was monitored throughout the evaluation to assess correlations between environmental humidity and hydrogel weight fluctuations.

[0146] Charge Injection Capacity (CIC) Measurement. The study characterized the electrochemical performance of GKAmp hydrogels using a three-electrode system to measure charge injection capacity (CIC). The working electrode consisted of a GKAmp hydrogel interface deposited on a 1.5 cm x 1.5 cm platinum (Pt) sheet. A Pt sheet of the same size served as the counter electrode, while an Ag / AgCl electrode immersed in saturated KCl solution served as the reference electrode. All electrodes were submerged in a 10 mM phosphate-buffered saline (PBS) electrolyte solution to maintain a consistent electrochemical environment. For CIC measurement, biphasic pulses of ± 0.5 V were applied at an interval of 0.01 s for 1500 cycles. The total charge injection densitywas determined using Equation 4._ Qa + Qcinj =(Eq. 4)

[0147] In Equation 4, Qaand Qcrepresent the cathodic and anodic charge quantities, respectively, and denote the effective hydrogel-electrode interface area. The CIC values recorded on the first day were normalized, and the long-term electrochemical stability was evaluated by repeating the test once per week for 4 weeks. A total of four independent samples were evaluated in each test to ensure reproducibility.

[0148] Cell Viability’ Test. The biocompatibility of the GK Amp was evaluated using a cytotoxicity assay based on the extract method. The GKAmp prepared with 3D-printed soft molds was immersed in Dulbecco's Modified Eagle Medium (DMEM) for 2 hours to obtain the hydrogel extract. A total of 1×105mouse NIH / 3T3 fibroblasts were seeded into a 96-well plate and incubated for 4 hours. When the cells were fully adhered and proliferating, the culture medium was replaced with the hydrogel extract solution. Cells cultured in standardAttorney Docket No. 10046-682W018684 WAN DMEM medium without hydrogel extract served as the control group. After incubation for 24 and 48 hours, cell viability was evaluated using the CCK-8 cytotoxicity assay kit according to the manufacturer’s protocol. Absorbance measurements were taken at 450 nm using a microplate reader. The relative cell viability was calculated using Equation 5.(Eq. 5)

[0149] In Equation 5, Asamplerepresents the absorbance of cells treated with the hydrogel extract, Acontrolis the absorbance of cells cultured in normal DMEM medium, and corresponds to the background absorbance.

[0150] Human Subjects / Participants and EEG Data Acquisition. Six healthy human subjects with normal or corrected-to-normal vision participated in the study (3 females, 3 males; age: 24.3 + 5.8 y). Electrodes were arranged according to the international 10-10 system, with FC3, FC4, FCz, Cl, C2, C3, C4, C5, Cz, CP3, and CP4 as recording electrodes. The reference electrode was positioned at CPz. This electrode arrangement was chosen to ensure optimal coverage of motor and sensorimotor cortical areas while maintaining high signal quality for motor imagery and rest-task analysis. All signals were collected using LabStreamingLayer (e.g., LSL; Swartz Center for Computational Neuroscience, University of California, San Diego, USA) to record in real time with a 32-channel research-grade EEG amplifier (e.g., ANT Neuro, Netherlands). Electrode impedance values were acquired using eego64 software (e.g., ANT Neuro, Netherlands).

[0151] One-Month Longitudinal Offline Motor Imagery EEG Recording. In each recording session (day), the subjects underwent four runs of motor imagery (MI) offline recordings using the experimental device. The scalp-electrode impedance measurements were taken at the end of each recording session. Additionally, a single control recording session using a commercial EEG system was conducted for each subject at an arbitrary time point within the 5 -week evaluation.

[0152] Due to the additional preparation time required for the commercial EEG system, each control recording session included 3 runs, whereas sessions for the experimental device included 4 runs per day. This adjustment ensured that the overall duration of each recording session remained comparable across systems, optimizing alignment in experimental conditions and minimizing potential variability in environmental factors or subjects’ fatigue.Attorney Docket No. 10046-682W018684 WAN

[0153] Interface for Offline Motor Imagery’ Task. The bar feedback task utilized in the MI experiment was derived from protocols adopted in BCI research over the past decade.Variants of the task have been implemented in previous studies

[0039] ,

[0049] ,

[0052] ,

[0053] ,

[0154] Each trial began with a 2-second fixation period, during which subjects were presented with a black screen containing white text instructing them to resume focus and prepare for the upcoming task. The 2-second fixation period was followed by a 2-second cue period, during which the type of task was indicated by the color of the oval icon: a blue oval signaled an upcoming rest task, while a red oval signaled an upcoming MI task. The cue period transitioned into a 6-second task period, during which subjects performed either a rest or an MI task. A running progress bar at the bottom of the screen provided real-time feedback, loading from 0% to 100% over 6 seconds. Each trial concluded with a 4-second intertrial relaxation period, allowing the subjects to reset before the subsequent trial.

[0155] For the rest tasks, a white cross overlaid the blue oval during the task period, visually reinforcing that the subjects should remain relaxed. For MI tasks, an arrow pointing toward the subjects’ dominant-hand side was overlaid on the red oval during the task period, prompting them to mentally rehearse the kinesthetic movements of their dominant hand. The visual feedback provided by this interface ensured that the subjects remained engaged and followed task instructions throughout each evaluation session.

[0156] EEG Signal Analysis and Visualization. Raw EEG signals were preprocessed using MATLAB. The preprocessing steps included spatial filtering, notch filtering, and bandpass filtering. During spatial filtering, the study used a Common Average Reference (CAR) spatial filter to reduce noise and enhance signal quality. During notch filtering, the study used a 60 Hz notch filter to remove powerline noise. During bandpass filtering, the study used a fourth-order Butterworth filter from 1 to 50 Hz to obtain the broadband EEG signal for early-stage visualization. The study then filtered signals of interest between 8—13 Hz (mu-band) or 18—30 Hz (beta-band) using a fourth-order Butterworth filter to retain task-relevant frequency bands.

[0157] Each trial included a fixation period, a cue period, and an assigned task, e.g., a motor imagery task or a rest task. All trials exceeding a maximum allowed amplitude threshold of 50 V during the evaluation period (last 2 seconds from fixation to the end of the task) were rejected to minimize contamination from artifacts such as muscle activity or eye movements. Valid trials were retained for further analysis. For each valid trial, feature extraction was conducted to isolate and analyze the mu (e.g., 8—13 Hz) and beta (e.g., 18-30 Hz) frequency bands. Bandpass filters were applied to the preprocessed EEG signals toAttorney Docket No. 10046-682W018684 WAN extract these frequency components. To quantify event-related desynchronization (ERD), power values during the task period were normalized relative to baseline periods corresponding to the fixation period. Next, grand averages of normalized power in the mu and beta bands were computed across trials for MI and rest tasks. Topographical maps of ERD in the mu and beta bands were created using MATLAB's EEGLAB toolbox. Data from all participants were averaged for each task (e.g., MI vs. rest) and plotted across electrode locations using standardized channel coordinates. Statistical comparisons between MI and rest tasks at different time points (e.g., Day 0, Day 4, Day 8, Day 15, Day 30) were performed using statistical methods disclosed herein.

[0158] Statistical Methods in EEG Analysis. The study employed a two-way repeated- measures analysis of variance (RM-ANOVA) to evaluate the longitudinal changes in motor imagery (MI) and rest-related EEG activity in the mu (e.g., 8—13 Hz) and beta (e.g., 18-30 Hz) frequency bands. The analysis considered two factors: task condition (e.g., MI vs. rest) and time (e.g., Day 0, Day 4, Day 8, Day 15, Day 30). The assumption of sphericity was tested and, where necessary, corrected using the Geisser-Greenhouse epsilon. Statistical significance was set at an alpha level of 0.05. The RM-ANOVA results identified effects for the column factor (e.g., task conditions) and interactions between task condition and time, indicating distinct differences in normalized power between MI and rest conditions across sessions. Post-hoc comparisons were conducted using Sfdak’s multiple comparisons test to evaluate pairwise differences between conditions at each time point and across time within each condition. Adjusted p-values were reported to control for family-wise error rates.

[0159] Signal quality metrics were compared between two systems: the experimental device and a control system (e.g., a commercial stretchable head cap with Ag / AgCl conductive gel electrodes). For both systems, trial rejection rates, RMS amplitudes, and composite metrics were evaluated across all rejection thresholds to evaluate noise levels and artifact tolerance.

[0160] Composite Metric Calculation. To evaluate overall signal quality within each evaluation session, a composite metric was calculated for each run using Equation 6.RMS AmplitudeComposite Metric — - - - (1 — Rejection Rate)(Eq. 6)

[0161] The composite metric integrated signal quality (e.g., RMS amplitude) with the rate for high signal quality trials (e.g., 1 -rejection rate), where lower values indicated better signal quality. All four runs from each evaluation session were analyzed in the experimentalAttorney Docket No. 10046-682W018684 WAN device to calculate channel-wise composite metrics across five rejection thresholds (e.g., 50 pV to 150 pV). To accommodate the sample size of the control system, the three runs with the lowest average composite metric values were selected for further analysis of direct comparison with the control system.

[0162] Trial Rejection Rate and RMS Amplitude Calculation. To evaluate signal quality, the study used a range of rejection thresholds (e.g., 50 pV, 75 pV, 100 pV, 125 pV, and 150 pV) to identify and exclude trials with excessive amplitude deviations. For each threshold, the maximum absolute value of each trial was calculated and compared with the threshold. RMS amplitudes were calculated for both MI and rest periods across all channels. Rejection rates were computed as the proportion of excluded epochs relative to the total number of epochs.

[0163] Online Closed-Loop Brain-Computer Interface (BCI) Evaluation. Subjects engaged in a motor imagery (MI) task to control a one-dimensional (ID) bar feedback system. In contrast to the one-month offline MI EEG recordings, the BCI experiment involved distinguishing MI of the dominant hand from rest using a subject-specific Riemannian geometry-based Minimum Distance to Mean (MDM) classifier to provide realtime feedback and BCI commands. Prior to the first online BCI session, subjects completed four offline calibration runs on the same day. Each run included multiple trials alternating between MI of the dominant hand and rest. The data collected during these runs was used to train the Riemannian MDM decoder.

[0164] The online experiment was conducted across three online BCI sessions. During the first online evaluation session, the subjects performed four online runs on the same day after the offline calibration. The second online evaluation session was conducted on the second day of the experiment and included four additional online runs. The third session took place 26 days later (±2 days), allowing a 4-week separation from the initial sessions.

[0165] To optimize user performance and ensure familiarity with the BCI system, subjects completed one practice run and one threshold-tuning run before each set of four online runs. The preparatory runs provided subjects with an opportunity to adapt to the BCI system and enabled adjustments to thresholds, ensuring a tailored user experience during the BCI experiments.

[0166] Feature Extraction. Signals were bandpass-filtered with a second-order Butterworth filter from 8 to 30 Hz to capture sensorimotor rhythms (SMRs). The filtered EEG data sample matrix, denoted as X ∈ RN×N, represents signals from Ncchannels over Nttime points. Covariance matrices, denoted as Mt, were computed for each 1 -second slidingAttorney Docket No. 10046-682W018684 WAN window using a shrinkage-based estimator (e.g., Ledoit and Wolf method) to enhance numerical stability. The covariance matrices were normalized by their trace to ensure they resided on the Riemannian manifold of symmetric positive definite (SPD) matrices, as shown in Equation 7.XXM,=________T______trace (XXT)(Eq. 7)

[0167] The normalization step in Equation 7 enabled the use of Riemannian geometry for classification, allowing the construction of a distance-based classifier that leveraged the manifold's structure to compute means and distances. The parametric equation of the shortest path, or geodesic, for curved spaces between two matrices M1and M2on the Riemannian manifold can be expressed parametrically as Equation 8.(Eq. 8)

[0168] Riemannian Geometry Framework. The study treated covariance matrices (denoted as M1and M2) derived from EEG signals as points on a Riemannian manifold, a smooth, differentiable space in which each point has an associated Euclidean tangent space. The study used the Affine-Invariant Riemannian Metric (AIRM) to compute distances between covariance matrices. For two matrices, the Riemannian distance could be defined per Equation 9, where A, are the eigenvaluesδr(M1, M2)(Eq. 9)

[0169] Class prototypes were calculated, using Equation 10, as the Riemannian mean (or Karcher mean) of all covariance matrices in a given class.(Eq. 10)

[0170] In Equation 10, Pnis the space of SPD matrices, and N is the number of covariance matrices in the class. The Minimum Distance to Mean (MDM) classifier assigned a test matrix Mtestto the class whose prototype had the smallest Riemannian distance, as shown in Equation 11, where Mkrepresents the prototype for the class k.Attorney Docket No. 10046-682W018684 WAN(Eq. 11)

[0171] BCI Controller / System. The BCI system estimated the probability distribution for each EEG sample to belong to the respective task classes (e.g., motor imagery vs. rest). To determine a command, the BCI system employed an evidence accumulation approach over time. Specifically, a 1 -second sliding window of EEG signals was processed every 62.5 ms; probabilities for each class were accumulated until one class's evidence exceeded a predefined threshold, triggering a command; and if no threshold was reached within a specified timeout period, the trial was classified as " Timeout," and no command was issued.

[0172] To ensure smooth evidence accumulation, the study used an exponential smoothing method shown in Equation 12.(Eq. 12)

[0173] In Equation 12, Probtis the accumulated probability for a given class at the ithsample, is the posterior probability of that class for the ithsample, and a is the smoothing factor. The study kept a = 0.05 to balance responsiveness and stability in real-time feedback.

[0174] Incremental Adaptation via Generic Recentering (GR). To account for intra-subject variability and temporal non-stationarities in EEG signals, the study implemented an unsupervised recentering mechanism after each online run, following the methods in [9], The recentering mechanism updated class prototypes on the Riemannian manifold using geodesic interpolation, as shown in Equation 13.(Eq. 13)

[0175] In Equation 13, Mnewis a new covariance matrix from online data,is the current class prototype, and rj is a learning rate controlling adaptation speed, which the study kept at g = 0.001 to be constant. The geodesic interpolation formula on the Riemannian manifold was expressed as shown in Equation 8. The GR mechanism ensured that prototype updates respected the geometric structure of the SPD manifold while allowing gradual adaptation to changes in neural activity patterns during prolonged sessions.

[0176] Online BCI Performance Analysis. To evaluate the performance of the online brain-computer interface (BCI), the study used three metrics, including sample-wiseAttorney Docket No. 10046-682W018684 WAN classification accuracy, bar dynamics, and BCI hit accuracy. Sample-wise classification, calculated per Equation 14, measured the percentage of individual samples correctly classified into their respective class (e.g., MI of the dominant hand vs. rest) during each run.(Eq. 14)

[0177] In Equation 14, TP, TN, FP and FN correspond to true positives, true negatives, false positives, and false negatives, respectively. Unlike trial-level performance, sample-wise accuracy was independent of thresholds used for evidence accumulation, providing a direct measure of classifier effectiveness.

[0178] The bar dynamics reflected the user's level of control over the BCI system during motor imagery (MI) task execution. The bar dynamics represented the proportion of time during a trial that the accumulated evidence from the BCI decoder supported the correct class. Bar dynamics were computed based on the percentage of samples where the accumulated probability for the correct class exceeded 50%, indicating successful control.

[0179] The BCI hit accuracy evaluated command delivery performance by calculating the ratio of trials with correct threshold hits to the total number of threshold crossings within a session. Trials resulting in a " Timeout" outcome were excluded from this calculation. The BCI hit accuracy was normalized to account for timeout trials, as shown in Equation 15, where ntimeoutsis the number of trials that ended with a timeout and ntrialsis the total number of trials in an evaluation session.(Eq. 15)

[0180] To establish baseline performance levels, chance levels for classification accuracy and command-delivery metrics were computed by randomly permuting the test labels 10,000 times. The outcomes were averaged over these permutations to determine the expected performance under random conditions.

[0181] Additional Discussion

[0182] Electroencephalography (EEG) has been used in neuroscience research and clinical applications for a century [1], Its temporal resolution enables real-time monitoring of neural dynamics, making it indispensable for diagnosing and managing neurological disorders, including epilepsy, sleep disorders, and traumatic brain injury (TBI) [2-4], In addition, EEG has emerged as a foundational technology for non-invasive brain-computerAttorney Docket No. 10046-682W018684 WAN interfaces (BCIs), facilitating direct communication between the brain and external devices without the need for a craniotomy. This advancement has broadened its utility beyond conventional medical contexts to encompass applications such as intelligent prosthetic control [5], [6], adaptive human-machine interaction [7-9], and neurorehabilitation training [10-12],

[0183] The EEG recording systems typically include a wired EEG head cap equipped with either wet or dry electrodes

[0013] . However, EEG head caps are rigid or exhibit limited stretchability. The solutions for EEG head caps either incorporate mechanisms to induce stretchability in the headset or offer a range of sizes

[0014] ,

[0015] , resulting in bulky systems or limiting the head caps' adaptability to various head shapes and sizes

[0016] . These constraints can increase operational costs for facilities engaged in EEG recordings

[0017] , particularly when catering to populations with diverse head dimensions, such as infants and adolescents, whose head sizes increase with development

[0018] , In addition to the adaptability of the head cap, the conundrum of recording in the hair-covered area remains unaddressed.

[0184] The current systems typically use either wet or dry Ag / AgCl electrodes

[0013] ,

[0019] , The wet-electrode configuration relies on conductive gels to create low -impedance connections with the scalp. Although wet electrodes provide high signal quality, they require preparation by trained personnel and cause discomfort for users due to the untidy nature of gel application and its propensity to dry out

[0020] . On the other hand, the dry counterpart eliminates the need for conductive gels while providing rapid setup and greater mobility

[0020] . However, dry electrodes often compromise overall signal quality due to poor electrode-skin contact during movement

[0021] . Dry electrodes often require a pointed or conical design to ensure contact with the scalp, particularly in areas with hair

[0020] , which can cause discomfort and make long-term monitoring impractical

[0020] , In addition to the above-discussed issues, hair complicates high-fidelity EEG recordings by obstructing the conformability of the head cap and the stable electrode-skin contact. In clinical practice, this issue is mitigated by shaving most of the subject’s hair to ensure reliable recordings, an impractical solution for everyday use that also imposes psychological burdens, particularly on children, adolescents, and female patients.

[0185] Conclusion

[0186] It is to be understood that the methods and systems are not limited to specific synthetic methods, specific components, or particular compositions. It is also to be understood that the terminology used herein is for the purpose of describing particular implementations only and is not intended to be limiting.Attorney Docket No. 10046-682W018684 WAN

[0187] As used in the specification and the appended claims, the singular forms “a,” “an” and “the” include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from “about” one particular value, and / or to “about” another particular value. When such a range is expressed, another implementation includes from one particular value and / or to another particular value. Similarly, when values are expressed as approximations, by use of the antecedent “about,” it will be understood that the particular value forms another implementation. It will be further understood that the endpoints of each of the ranges are significant both in relation to the other endpoint and independently of the other endpoint.

[0188] “Optional” or “optionally” means that the subsequently described event or circumstance may or may not occur and that the description includes instances where said event or circumstance occurs and instances where it does not.

[0189] Throughout the description and claims of this specification, the word “comprise” and variations of the word, such as “comprising” and “comprises,” mean “including but not limited to,” and are not intended to exclude, for example, other additives, components, integers, or steps. “Exemplary” means “an example of’ and is not intended to convey an indication of a preferred or ideal implementation. “Such as” is not used in a restrictive sense but for explanatory purposes.

[0190] Disclosed are components that can be used to perform the disclosed methods and systems. These and other components are disclosed herein, and it is understood that when combinations, subsets, interactions, groups, etc. of these components are disclosed while specific reference of each various individual and collective combinations and permutation of these may not be explicitly disclosed, each is specifically contemplated and described herein, for all methods and systems. This applies to all aspects of this application, including, but not limited to, steps in disclosed methods. Thus, if there are a variety of additional steps that can be performed it is understood that each of these additional steps can be performed with any specific implementation or combination of implementations of the disclosed methods.

[0191] Compounds are described using standard nomenclature. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as is commonly understood by one of skill in the art to which this disclosure belongs.

[0192] As used herein, a “hydrogel” is a three-dimensional, crosslinked network of hydrophilic polymer chains that is insoluble in water and capable of absorbing and retaining an aqueous fluid such that the network is swollen throughout its volume. An “ionic hydrogel” is a hydrogel in whose polymer chains bear ionizable or permanently charged groups alongAttorney Docket No. 10046-682W018684 WAN the polymer backbone and / or pendant chains such that an overall ionic character is imparted to the swollen network.

[0193] The following patents, applications, and publications, as listed below and throughout tills document, are hereby incorporated by reference in their entirety herein. 1. Arjoonsingh, A., Jamal, B. C. & Ganti, L. History and evolution of the electroencephalogram. Cureus 16, e66385 (2024).2. Munia, T. T. K., Haider, A., Schneider, C., Romanick, M. & Fazel-Rezai, R. A novel EEG based spectral analysis of persistent brain function alteration in athletes with concussion history. Sci. Rep. 7, 17221 (2017).3. Liu, M., Liu, B., Ye, Z. & Wu, D. Bibliometric analysis of electroencephalogram research in mild cognitive impairment from 2005 to 2022. Front. Neurosci. 17, 1128851 (2023).4. Nuwer, M. R., Hovda, D. A., Schrader, L. M. & Vespa, P. M. Routine and quantitative EEG in mild traumatic brain injury. Clin. Neurophysiol. 116, 2001-2025 (2005).5. Beyrouthy, I’., Al Kork, S. K., Korbane, J. A. & Abdulmonem, A. EEG Mind controlled Smart Prosthetic Arm. in 2016 IEEE International Conference on Emerging Technologies and Innovative Business Practices for the Transformation of Societies (EmergiTech) 404 -409 (IEEE, 2016).6. Al-Quraishi, M. S., Elamvazuthi, I., Daud, S., Parasuraman, S. & Borboni, A. EEGbased control for upper and lower limb exoskeletons and prostheses: A systematic review. Sensors (Basel) 18, (2018).7. Stephygraph, L. R. & Arunkumar, N. Brain-actuated wireless mobile robot control through an adaptive human-machine interface, in Proceedings of the International Conference on Soft Computing Systems 537-549 (Springer India, New Delhi, 2016). 8. Zhang, J., Yin, Z. & Wang, R. Design of an adaptive human-machine system based on dynamical pattern recognition of cognitive task-load. Front. Neurosci. 11, 129 (2017).9. Kumar, S., Alawieh, H., Racz, F. S., Faklireddine, R. & Millan, J. D. R. Transfer learning promotes acquisition of individual BCI skills. PNAS Nexus 3, gae076 (2024).10. Ang, K. K. et al. Clinical study of neurorehabilitation in stroke using EEG-based motor imagery brain-computer interface with robotic feedback. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 2010, 5549-5552 (2010).11. Millan, J. D. R. et al. Combining brain-computer interfaces and assistive technologies:State-of-the-art and challenges. Front. Neurosci. 4, (2010).12. Biasiucci, A. et al. Brain-actuated functional electrical stimulation elicits lasting armAttorney Docket No. 10046-682W018684 WAN motor recovery after stroke. Nat. Commun. 9, 2421 (2018).Casson, A. J. Wearable EEG and beyond. Biomed. Eng. Lett. 9, 53-71 (2019).Soufineyestani, M., Dowling, D. & Khan, A. A. Electroencephalography (EEG) technology applications and available devices. Appl. Sci. (Basel) (2020) doi:10.3390 / appl0217453.Sawangjai, P., Hompoonsup, S., Leelaarporn, P., Kongwudhikunakorn, S. & Wilaiprasitporn, T. Consumer grade EEG measuring sensors as research tools: A review. IEEE Sens. J. 20, 3996-4024 (2020).David Hairston, W. et al. Usability of four commercially-oriented EEG systems. J. Neural Eng. 11, 046018 (2014).Shields, S. M., Morse, C. E., Applebaugh, E. D., Muntz, T. L. & Nichols, D. F. Are electrode caps worth the investment? An evaluation of EEG methods in undergraduate neuroscience laboratory courses and researchJ. Undergrad. Neurosci. Educ. 15, A29-A37 (2016).Farkas, L. G., Posnick, J. C. & Hreczko, T. M. Anthropometric growth study of the head. Cleft Palate Craniofac. J. 29, 303-308 (1992).Hsieh, J. C., Alawieh, H., Millan, J. D. R. & Wang, H. The evolving landscape of non- invasive EEG-brain-computer interfaces: Current trends and future perspectives. Device 2, (2024).Lopez-Gordo, M. A., Sanchez-Morillo, D. & Pelayo Valle, F. Dry EEG electrodes. Sensors 14, 12847-12870 (2014).Hsieh, J.-C. et al. Design of hydrogel-based wearable EEG electrodes for medical applications. J. Mater. Chem. B Mater. Biol. Med. (2022) doi:10.1039 / d2tb00618a. Tian, L. et al. Large-area MRI-compatible epidermal electronic interfaces for prosthetic control and cognitive monitoring. Nat. Biomed. Eng. 3, 194-205 (2019).Scalco de Vasconcelos, L. et al. On-scalp printing of personalized electroencephalography e-tattoos. Cell Biomaterials 100004 (2099).Wang, C. et al. On-skin paintable biogel for long-term high-fidelity electroencephalogram recording. Sci Adv 8, eabo1396 (2022).Luo. J. et al. MXene-Enabled Self-Adaptive Hydrogel Interface for Active Electroencephalogram Interactions. ACS Nano 16, 19373-19384 (2022).Hsieh, J.-C. et al. Design of an injectable, self-adhesive, and highly stable hydrogel electrode for sleep recording. Device (2023).Rogers, J. A., Someya, T. & Huang, Y. Materials and mechanics for stretchableAttorney Docket No. 10046-682W018684 WAN electronics. Science 327, 1603-1607 (2010).Shyu, T. C. et al. A kirigami approach to engineering elasticity in nanocomposites through patterned defects. Nat. Mater. 14, 785-789 (2015).Zhao, R., Lin, S., Yuk, H. & Zhao, X. Kirigami enhances film adhesion. Soft Matter 14, 2515-2525 (2018).Zhang, Y. et al. Printing, folding and assembly methods for forming 3D mesostructures in advanced materials. Nat. Rev. Mater. 2, 1-17 (2017).Mahmood, M. et al. Fully portable and wireless universal brain-machine interfaces enabled by flexible scalp electronics and deep learning algorithm. Nat. Mach. Intell. 1, 412-422 (2019).Haykin, S. Communication systems. (2008).Cameron, J. & Skabara, P. J. The damaging effects of the acidity in PEDOT: PSS on semiconductor device performance and solutions based on non-acidic alternatives.Mater. Horiz. 7, 1759-1772 (2020).Hsieh, J.-C. et al. A highly stable electrode with low electrode-skin impedance for wearable brain-computer interface. Biosens. Bioelectron. 218, 114756 (2022).Bhattacharyya, S. & De, S. Gel electrophoresis and size selectivity of charged colloidal particles in a charged hydrogel medium. Chem. Eng. Sci. 141, 304-314 (2016).Xue, H. et al. Hydrogel electrodes with conductive and substrate-adhesive layers for noninvasive long-term EEG acquisition. Microsyst. Nanoeng. 9, 79 (2023).Vasilyev, A., Liburkina, S., Yakovlev, L., Perepelkina, O. & Kaplan, A. Assessing motor imagery in brain-computer interface training: Psychological and neurophysiological correlates. Neuropsychologia 97, 56-65 (2017).Liao, W., Li, J., Zhang, X. & Li, C. Motor imagery brain-computer interface rehabilitation system enhances upper limb performance and improves brain activity in stroke patients: A clinical study. Front. Hum. Neurosci. 17, 1117670 (2023).Alawieh, II. et al. Transcutaneous electrical spinal cord stimulation promotes focal sensorimotor activation that accelerates brain-computer interface skill learning. medRxiv (2024) doi:10.1101 / 2024.06.10.24308723.Fitts, P. M. & Posner, M. I. Human Performance. (Praeger, Westport, CT, 1979).Neuper, C., Wörtz, M. & Pfurtscheller, G. ERD / ERS patterns reflecting sensorimotor activation and deactivation. Prog. Brain Res. 159, 211-222 (2006).Vidaurre, C. & Blankertz, B. Towards a cure for BCI illiteracy. Brain Topogr. 23, 194-198 (2010).Attorney Docket No. 10046-682W018684 WAN Kilavik, B. E., Zaepffel, M., Brovelli, A., MacKay, W. A. & Riehie, A. The ups and downs of β oscillations in sensorimotor cortex. Exp. Neurol. 245, 15-26 (2013).Pfurtscheller, G., Neuper, C., Brunner, C. & da Silva, F. L. Beta rebound after different types of motor imagery in man. Neurosci. Lett. 378, 156-159 (2005).Neuper, C. & Pfurtscheller, G. Event-related dynamics of cortical rhythms: frequency¬ specific features and functional correlates. Int. J. Psychophysiol. 43, 41-58 (2001). McFarland, D. J. & Wolpaw, J. R. Sensorimotor rhythm-based brain-computer interface (BCI): model order selection for autoregressive spectral analysis. J. Neural Eng. 5, 155- 162 (2008).Yuan, H. & He, B. Brain-computer interfaces using sensorimotor rhythms: current state and future perspectives. IEEE Trans. Biomed. Eng. 61, 1425-1435 (2014).Oliveira, A. S., Schlink, B. R., Hairston, W. D., Konig, P. & Ferris, D. P. Proposing metrics for benchmarking novel EEG technologies towards real-world measurements. Front. Hum. Neurosci. 10, 188 (2016).Racz, F. S. et al. Combining detrended cross-correlation analysis with Riemannian geometry-based classification for improved brain-computer interface performance. Front. Neurosci. 18, 1271831 (2024).Hsieh, J.-C., Yao, M., Baird, B. & Wang, H. Protocol to fabricate a self-adhesive and long-term stable hydrogel for sleep EEG recording. STAR Protoc. 5, 103097 (2024). Kothe, C. Lab streaming layer (LSL). 26, 2015 (2014).Neuper, C., Scherer, R., Wriessnegger, S. & Pfurtscheller, G. Motor imagery and action observation: modulation of sensorimotor brain rhythms during mental control of a braincomputer interface. Clin. Neurophysiol. 120, 239-247 (2009).Leeb, R. et al. Transferring brain-computer interfaces beyond the laboratory: successful application control for motor-disabled users. Artif. Intell. Med. 59, 121-132 (2013).

Claims

Attorney Docket No. 10046-682W018684 WAN What is claimed:

1. A wearable device for measuring electroencephalography (EEG), the wearable device comprising:a stretchable substrate configured to conform to a head region of a user;an interconnect layer comprising a stretchable electrode array, formed in or on the stretchable substrate, comprising a plurality of conical electrodes configured to conform to the head region and measure EEG signals therefrom, wherein each of the plurality of conical electrodes comprises an ionic hydrogel; anda controller operatively coupled to the stretchable electrode array, wherein the controller is configured to receive and output the measured EEG signals, wherein the output measured EEG signals are subsequently used for real-time monitoring of a brain activity of the user for a period of time.

2. The wearable device of claim 1, wherein the interconnect layer further comprises: a set of stretchable interconnects electrically connecting the plurality of conical electrodes,wherein the set of stretchable interconnects is configured to maintain an attenuation value of the measured EEG signals below a predefined threshold attenuation value in response to a deformation of the stretchable electrode array.

3. The wearable device of claim 2, wherein the predefined threshold attenuation value is 0.2 decibels (dB).

4. The wearable device of any one of claims 1-3, wherein the controller is further configured to:adjust, via an amplifier or a filter, the measured EEG signals; andstore, in a data storage, the adjusted measured EEG signals, wherein the stored measured EEG signals are subsequently used for offline or longitudinal analysis of the brain activity.

5. The wearable device of any one of claims 1-4, wherein the stretchable substrate is configured with a Kirigami pattern configured to be in a first Kirigami configuration when placed on the head region and in a second Kirigami configuration when stretched from the first Kirigami configuration.Attorney Docket No. 10046-682W018684 WAN6. The wearable device of any one of claims 1-5, wherein the ionic hydrogel of each conical electrode is configured to penetrate hair layers and conformably contact skin of the head region to measure EEG signals therefrom.

7. The wearable device of any one of claims 1-6, wherein each of the plurality of conical electrodes defines an outer shell and comprises:a top region;a porous central region; anda bottom region having a smaller surface area than the top region and comprising the ionic hydrogel, wherein the bottom region is configured to make contact with a subject’s scalp.

8. The wearable device of claim 7, wherein each ionic hydrogel is at least partially removable and / or wherein each ionic hydrogel is configured to at least partially move within the outer shell when the respective conical electrode is in contact with the subject’s scalp.

9. The wearable device of any one of claims 1-8, wherein the stretchable substrate comprises a plurality of regions, including a temporal region having a first predefined pattern and a scalp region having a second predefined pattern.

10. The wearable device of claim 9, wherein the temporal region comprises a first plurality of apertures featuring a first horizontal spacing and first vertical spacing in relation to one another, wherein the scalp region comprises a second plurality of apertures featuring a second horizontal spacing and a second vertical spacing in relation to one another that is different from the first horizontal spacing and the first vertical spacing such that the temporal region has a higher aspect ratio than the scalp region.

11. The wearable device of any one of claims 1-10, wherein the stretchable substrate comprises a plurality of apertures to facilitate expansion and deformation of the substrate to adapt to different head sizes.

12. The wearable device of claim 11, wherein the plurality of apertures comprises slashes, lines, circles, ovals and / or geometric shapes.Attorney Docket No. 10046-682W018684 WAN 13. The wearable device of any one of claims 2-12, wherein the set of stretchable interconnects is coupled to the plurality of conical electrodes via a plurality of snap connectors.

14. The wearable device of claim 13, wherein each of the plurality of snap connectors comprises a male snap connector attached to the respective conical electrode and a corresponding female snap connector attached to a channel location of the interconnect layer.

15. The wearable device of claim 6, wherein: the ionic hydrogel comprises glycerol, potassium chloride (KCl), and 2-acrylamido-2-methyl-1-propanesulfonic acid sodium salt (AMPSNa), each stretchable interconnect comprises a liquid eutectic gallium-indium material (EGain), and the stretchable substrate comprises at least a silicone elastomer.

16. The wearable device of any one of claims 1-15, wherein the wearable device is used for monitoring and / or treatment of at least one of:Parkinson’s disease, epilepsy, Alzheimer’s disease, stroke, traumatic brain injury, psychiatric disorders, pain, peripheral chronic joint pain, overactive bladder syndrome, sleep disorders, carpal tunnel syndrome, visual prosthetics and / or blindness, mood disorders, or cognitive / neurodevelopmental disorders.

17. A method of fabricating a wearable device for measuring electroencephalography (EEG), the method comprising:providing a stretchable substrate in a substrate mold;embedding a Kirigami mesh template into the substrate mold to define a structural layout within the provided stretchable substrate;removing the embedded Kirigami mesh template from the substrate mold in response to the provided stretchable substrate being cured;placing a stretchable electrode array comprising a plurality of conical electrodes on or within the provided stretchable substrate corresponding to the structural layout; and depositing a set of stretchable interconnects into the provided stretchable substrate to electrically connect the plurality of conical electrodes of the stretchable electrode array.

18. The method of claim 17, wherein the set of stretchable interconnects forms an interconnect layer within the provided stretchable substrate to define conductive pathways between the plurality of conical electrodes.Attorney Docket No. 10046-682W018684 WAN19. The method of claim 17 or 18, further comprising:curing an ionic hydrogel within each of the plurality of conical electrodes prior to placing the plurality of conical electrodes on or within the provided stretchable substrate.

20. The method of any one of claims 17-19, further comprising:encapsulating the wearable device with a protective layer.

Images