Flexible brain electronic sensors advance wearable brain-computer interface

Abstract

The emerging field of wearable brain-computer interface (BCI) strives to achieve both high spatial and temporal resolution. The performance of flexible brain electronic sensor (FBES) has been validated across a variety of experimental settings, demonstrating their potential for real-world applications. As a result, FBES are increasingly shaping the landscape of health monitoring and disease treatment by enabling non-invasive, precise neural data acquisition. This review summarizes recent studies recent progress in wearable brain computer interface technology and FBES development, while provides insights into future clinical application of FBES within BCI systems. Additionally, we propose strategic directions to bridge the gap between laboratory research and practical healthcare implementations.

Introduction

The concept of brain computer interface (BCI) was introduced in 1924. BCI establishes a valid connection between the human or animal brain and a computer system, facilitating the exchange and transmission of information. Over the past century, BCI has evolved through stages of conceptual proof, theoretical development, application enhancement, and widespread adoption across various sectors, including medical, healthcare, smart home technologies, and specialized industries^1–5. Currently, medical treatment stands as a primary application area for BCI, serving as an effective tool in disease screening⁶, diagnosis⁷, and treatment⁸, Fig. 1 illustrates the working mechanism of BCI. The rapid development of wearable devices that are driven by the portability and sustainable monitoring advantages in recent years exerts significant influences on the progression of BCI technology towards improved portability^9–11. The advancements of wearable BCI technology have been leveraged in trials and clinical settings^12,13. Noteworthy, BCI technology features prominently in the strategic development plans of numerous countries worldwide. For instance, projects like the “BCI based human-computer interaction” sponsored by the national science foundation and the establishment of EBRAINS, an infrastructure for brain research by the European Union in 2023, illustrate the growing importance of BCI in major initiatives like science and technology innovation 2030 project that focuses on brain science, brain-like research, and the development of the next generation of artificial intelligence. It is evident that wearable BCI represents a crucial avenue for emerging technologies in the future.

Fig. 1.

The structure of BCI.

The brain electronic sensors serve as crucial components in BCI, facilitating brain signal acquisition and bidirectional transmission in wearable BCI systems. The performance of brain electronic sensors is contingent upon the used material, structure, and manufacturing processes. Traditional rigid sensors exhibit poor tensile, bending, and fatigue resistance, thus encountering consequential risk of brain damage upon implantation. To address these challenges, previous work focused on material synthesis and structural optimization efforts¹⁴, yet struggle to fully align with the demands of clinical medicine. In recent years, the development of flexible materials has revolutionized brain electronic sensors. FBES have emerged as a favorable option for wearable BCI, featuring superior flexibility and robust biocompatibility, which enables continuous monitoring of brain vital signs^15–17. The strength of brain signals is weak, for example, electroencephalography signals are measured to be merely 10–50 μV and magnetoencephalogram signals originating of synchronized activity of 10⁵ neurons show about 100fT in intensity^18,19. Therefore, the acquisition and anti-noise capabilities of FBES have higher requirements. The FBES applied in wearable BCI is confronted with many difficulties and concerns for scholars. With the development of flexible electronics, progressive enhancements in the performance and function of FBES have been achieved through material synthesis, structural innovations, and resolution of the disparate elastic modulus issues between metals and human tissues. This progress allows for multidirectional, multidimensional, and multilevel monitoring and collection of human or animal physiological signals, making them an ideal choice for wearable BCI devices, such as electronic skins, heart monitors, and brain ultrasound patches^20–22. Smith et al. discussed the utilization of flexible electronic devices in the realms of neurosurgery, soft surgical instruments, soft drug delivery devices and soft neural electrodes²³. Despite significant innovations and achievements, lingering technical challenges persist, including poor coupling between the FBES and skin, limited anti-interference capabilities of brain patches, instability of continuous working of composite flexible materials, and signal attenuation by the skull²⁴. Solving the technical bottleneck of FBES is the core of the development of BCI²⁵. Recently, Men et al. reviewed the research on flexible wearable electronic sensors²⁶, moreover, the application of flexible bioelectronics in personalized health management and nerve disease treatment were discussed^27,28. Nevertheless, there remains a notable absence of discourse regarding the advancements in research concerning progress of flexible brain electronic sensor based on wearable BCI. Recent advances in FBES include the expansion of signal acquisition methods, the diversification of sensing principles, and the innovation of materials and the optimization of device structure (Fig. 2).

Fig. 2.

The summarize of FBES of sensing classification, structure and types of signal acquisition in brain.

Based on the above discussion, this review focuses on the application of BCI in neurological diseases, scrutinizes various types and methodologies of brain signal data collection, summarizes the sensing mechanisms and characteristics of FBES, reviews the materials and structures used in FBES. Finally, the challenges faced in the development of BCI technology and FBES and the outlook for future development and application are discussed, wherein the next research hotspot or direction will focus on reducing power consumption, optimizing microprocessor performance, implementing machine learning, and exploring multimodal information parallel sampling for further exploration. These achievements will expedite the utilization of wearable BCI technology based on FBES in the realms of brain disease diagnosis, treatment and rehabilitation.

Research status of wearable BCI

Traditional brain monitoring equipment can diagnose brain tumors, thrombosis, Intracranial pointer, etc. Commonly-used diagnostic tools include digital radiography²⁹, computed tomography³⁰, magnetic resonance imaging^31,32, and neuro-electrophysiological apparatus³³. Although the kinds of equipment technology are plentiful and various, their volume is large. In recent years, the development of portable micro monitoring equipment has become a hot field. For example, small intelligent mobile computed tomography robots (Fig. 3)³⁴.

Fig. 3.

The moving CT of small intelligent robot. Reproduced with permission³⁴, Copyright Sichuan Provincial People’s Hospital.

The BCI assisted disease diagnosis

Current advancements in brain science research primarily focus on employing portable BCI equipment to continuously monitor brain neurons, neural activities, and cerebrovascular functions to facilitate disease diagnosis. Motor imagery has emerged as a predominant therapeutic paradigm in BCI applications. Typically, Zhang et al. utilized functional near-infrared spectroscopy to capture motor imagery-related information, wherein they explored the validity of such a common spatial pattern method based on functional near-infrared spectroscopy for motor imagery BCI³⁵. Wireless transmission represents a cutting-edge technology in the BCI domain. He et al. reviewed the development trends in wearable and wireless electroencephalogram systems in the past decade, compared the application and market situation of the wireless BCI system, the portable wireless electroencephalogram systems had three main application areas: consumer, clinical and research³⁶. Additionally, Mahmood et al. unveiled a virtual-reality portable BCI enabled by wireless soft bioelectronics³⁷. With continuous enhancements in hardware and software, the future trajectory of wearable BCI systems for data acquisition and control is likely to pivot towards wireless transmission methods (Fig. 4a).

Fig. 4. Wearable BCI applications in diagnosis, health management, and treatment.

a The tester wears an image of the VR soft system, including the hardware system (left), and an example of text spelling using EEG (right). Reproduced with permission³⁷, Copyright 2022, Elsevier. b Structural diagram of a triboelectric ear-mounted sensor integrated with oect. Reproduced with permission³⁸, Copyright 2024, Elsevier Ltd. c The hypnotic waveform of the test subjects throughout the night. The solid blue line represents the true sleepgram, and the dashed red line predicts the model sleepgram⁴¹. Copyright 2023, Liu et al. d Design of the SLEEPFAST study⁴³ Copyright 2024 Bressler et al. e An optical flow device was implanted under the skin of the liptodonte skull. Reproduced with permission⁵², Copyright 2018, Wiley-VCH.

Acquiring brain signals involves implanting sensors on the surface or inside the brain. There is a severe attenuation as it passes through the skull. In-ear sensors facilitate inconspicuous brain activity monitoring, so acquiring brain signals by the cochlea is a viable strategy. With considering comfort and connectivity, Wei et al. reported on an ear-worn triboelectric sensor, enables continuous monitoring of facial expressions (Fig. 4b)³⁸. Further, dual-modal wearable BCI based on flexible sensors in the ear and visual stimulation have been investigated. Xue et al. proposed a visual and auditory BCI system based on in-ear bioelectronics, which can expand spirally along the auditory canal by electrothermal drive. Subject steady-state visual evoked potential (SSVEP) achieved 95% off-line accuracy in BCI classification, which indicates that the auricular perception study is an effective means to assist harmonics in the spatial distribution of SSVEP³⁹. The ear canal was close to the central nervous system, it was an effective way to collect brain signs through the ear canal. The in-ear integrated array of electrochemical and electrophysiological sensors was investigated by Gert, it was positioned on a flexible substrate around the headset, and the lactate concentration and brain status could be monitored⁴⁰. With the advances in materials science, the FBES is being explored which can penetrate the skull window, such as through the eye, ear canal etc.

The wearable BCI facilitates health management

Wearable BCI has potential applications in health management, e.g., sleep monitoring using wearable BCI devices offers a valuable approach for managing sleep disorders. Traditionally, sleep monitoring had been confined to hospital settings, but the advent of portable wearable sleep monitoring systems can overcome this limitation. The portability concern could be well solved by the wearable sleep monitoring system. In a notable study, Liu et al. introduced a wearable home insomnia management system that underlined a body motion recorder for the qualitative assessment of insomnia and the evaluation of insomnia treatment outcomes (Fig. 4c)⁴¹. Additionally, polysomnography represents a reliable method for evaluating and managing sleep issues⁴². Bressler et al. conducted a multi-night randomized controlled crossover trial employing wearable neuromodulation devices on adults with sleep onset latencies over 30 min. Their results indicated that the intervention could serve as an alternative to pharmacological treatments and preferably extend sleep duration for individuals (Fig. 4d)⁴³. In the future, the application interface of wearable BCI technology will be further expanded, such as postoperative monitoring, health data collection and so on.

Auxiliary diseases continue to treat

Portable BCI is an effective way to assist disease treatment. Portable BCI provided a communication way for those who had difficulty in speaking or moving, such as locked-in syndrome or amyotrophic lateral sclerosis. However, reactive BCIs like P300 and SSVEP need stimulation from external devices⁴⁴, which affected the fatigue resistance of BCI. The BCI of motor imagination negates the need for simulation, but it is constrained when used to control a computer. A wearable ear-EEG acquisition tool was developed by Kaongoen N. and the developed experiment showed no significant difference between the performance of ear-EEG and scalp-EEG⁴⁵. Electrocorticogram is often used to describe epileptic regions of brain and assist in surgical removal. Stephanie P. Lacour worked on a flexible electrode array for manufacturing large-area robots, and flexible robotic electrode arrays were deployed on the cortex of miniature pigs to monitor cortical activity⁴⁶. This breakthrough provides new ideas for minimally invasive cortical surgery and treatment related to neurological diseases.

The challenges of FBES in BCI applications

As the core device of brain-computer interface, flexible brain electronic sensors affect the whole system, and thus it is demanded to meet some challenges in the process of sensing application. For example, acute and chronic mechanical injury of FBES and insufficient detection depth limit their clinical application^47,48. In particular, shielding and signal attenuation induced by skull are the core challenges that restrict the applications of FBES in BCI. The electrical conductivity of skull and scalp tissue is quite different (the electrical conductivity of skull is about 0.01–0.02 S/m, and that of scalp conductivity is about 0.1-0.3 S/m), resulting in the attenuation of electrical signal up to 80–90%. In particular, the low-frequency signal attenuation is the most prominent, such as Delta and Theta waves. Though invasive flexible electrodes offer direct contact with neurons, the bottlenecks such as poor compatibility with biological tissues and insufficient long-term stability remain existent to limit the application of wearable BCI. The attenuation coefficient of ultrasound signal to the skull is about approximately 1.3 dB/cm·MHz, and the energy loss of high frequency ultrasound ( > 1 MHz) after penetrating through the skull is up to 70%, resulting in weak signal echoes, while low frequency ultrasound ( < 1 MHz) has strong penetration but low imaging resolution. To break the technical bottleneck of signal attenuation by the skull, flexible electrodes and conformal devices have been developed to reduce signal attenuation through optimally fitting the ear canal to support visual and auditory South Pole interface applications³⁹. Additionally, combining quantum coding with deep learning to suppress the influence of skull acoustic heterogeneity can improve the depth of stroke monitoring⁴⁹. Beyond them, metamaterial and plane wave technology, high frequency transmission and low frequency reception strategy are also conducive to improving signal attenuation⁵⁰. In the future, multi-mode fusion with electrical, acoustic, magnetic and optical imaging are expected to reduce these limitations especially after combining with flexible and degradable electrodes and the signal complementarity capable of enhancing penetration and resolution. The FBES implanted in biological tissues need to consider biocompatibility in case of biological rejection. However, the existing flexible electronics technology cannot completely solve the problem. Typically, the monitoring performance of FBES still cannot fully meet the needs of clinical diagnosis and treatment⁵¹, such as low drug delivery efficiency by a streamer (Fig. 4e)⁵². These problems still need to be addressed by the interdisciplinary integration of electronics, materials and medicine.

The type and method in signal acquisition

Currently, the composition of BCI roughly includes brain signal acquisition, signal transmission and transformation, information processing and analysis and control, no matter what whether it is implantable or non-implantable portable BCI. Implantable FBES are typically inserted through craniotomy for direct contact with areas like the cerebral infarction module and cerebral cortex surface. FBES is a tool for the diagnosis and treatment of brain diseases. Nerve stimulators, cochlear implants and spinal cord stimulators are common invasive flexible brain electronic sensors. In order to evaluate the biocompatibility of FBES, many jobs have been done, such as material toxicity experiments⁵³, and sensor fabrication also deserves to comprehensively considered so as to match the size of biological tissues (Fig. 5c)^54,55.

Fig. 5. Microprobe and brain tissue sign monitoring experiment.

a Schematic of the light probe implanted in brain tissue (left: normal; Right: low oxygen levels)⁵⁹ Copyright 2024, Cai et al. b Schematic representation of wrist rehabilitation based on visually guided attentional brain control. Reproduced with permission¹³⁵, Copyright 2019, IEEE. c Wireless communication between magnetic implants and BCI⁵⁵ Copyright 2024, Wan et al. d Left: Brain lactate changes were recorded during exercise. Right: Asterisks indicate the time of onset of sweating⁴⁰ Copyright 2024, Xu et al.

The implantable FBES

Implantable FBES inevitably causes scalp damages. To reduce the risk of craniotomy, Kenneth L. Shepard et al. studied a surface electrode. The electrodes could selectively shape charge injection at a depth of 300 microns with a lateral resolution beyond 100 microns⁵⁶. In addition, structural innovation was also an ideal route to reduce the risk of implanting sensors, e.g., Zhang et al. designed an ultrasmall and flexible endovascular neural probe⁵⁷. Many factors should be considered after FBES implantation. The steady-state change and nonlinearity of FBES should be considered when the probe was electrically stimulated to the brain. In addition, the chemical or biochemical factors of the implantable FBES should also be considered⁵⁸. Matching the mechanical and biological properties of implantable FBES is an effective way to reduce the postoperative immune response. The size of FBES should be tailored to individual neurons or axons, under which the FBES can be seamlessly integrated with neural tissue after being implanted.

The ethical concerns in terms of human implantable BCI remain a contentious issue, and some experimental animals with FBES implanted in their skulls are subjected to many sufferings such as lose control and seizures. Implanted FBES-asired Infection and immune rejection after surgery as well as the use of biological information (neural signals, physiological data) for non-medical purposes are also ethical issues of concern. BCI companies, such as neuralink, have been sued by animal rights groups. Fortunately, in 2021, Blackrock Neurotech’s MATM BCI system received FDA approval as a medical device. Zhao conducted a study of implantable photoprobes, and their developed probes continuously monitored the movement of PbtO2 signals in deep brain through a wireless connection (refer to Fig. 5a)⁵⁹. Afterwards, Zhao’s team implanted BCI processors with the size of two coins into the skull of a 54-year-old paraplegic patient, and currently the clinical trials of minimally-invasive implantable BCI had been completed⁶⁰. Despite the completion of these trials, the approval for human clinical trials of implanted BCI technology is still pending due to uncertainties regarding surgical risks, safety, and ethical concerns. Consequently, researchers are primarily focusing on the development of non-implantable flexible sensors for monitoring brain activity and vital signs.

The non-implantable FBES

The non-implantable FBES is applied to the surface of brain tissue to collect brain signs. FBES has the advantages of high safety and convenience. The signals collected by non-implantable FBES include temperature, electroencephalogram and sweat lactic acid⁶¹, sweat⁴⁰, flexible optogenetics^62,63, wireless energy transfer^64,65, etc. The biosafety of non-implantable FBES is ideal, but the coupling of FBES to the head epidermis remains a challenge (Fig. 5d).

Non-implantable FBES based on temperature are early brain signal detectors. The flexible brain temperature sensor should be simple in principle, thin and easy to be installed, which dictates a good choice for portable BCI. Currently, the implanted flexible brain temperature sensors have been widely used in wearable BCI. Zhao et al. systematically studied the influences of material selection, structure design and process improvement on the performance of multifunctional flexible temperature sensors⁶⁶. Due to the attenuation effect of skull, the spatiotemporal accuracy of signals collected by non-implantable FBES is limited. New materials and new technologies also cater to higher requirements. As a paradigm, miniaturized electrocorticography electrode arrays based on microelectromechanical system technology increase complaisant compliance. The electrical signals of neurons are in the microvolt level, so the detection accuracy of the sensor is a challenge. Zhou et al. detected electroencephalogram signals of neurons and blood vessel blood sample signals by non-invasive method⁶⁷. Xu had made a number of achievements in non-implantable flexible patches, such as portable echocardiographic patches²¹, ears micro sensor⁶¹. Wu et al. proposed flexible optogenetic device that could achieve high-quality neural signal recording and significantly enhance neural activity after light stimulation⁶². Due to the lack of contact with brain nerves and cerebral vessels, non-implantable FBES is limited in the acquisition of deep brain signals, and there is still broad room for improvement in the diagnosis of complex brain diseases. The picture of the structure and outline of the non-implanted FBES is shown in Fig. 6.

Fig. 6. The structural and contour drawings of the non-implantable FBES.

a The Structure of a wearable multifunctional sensor for the detection of forehead EEG signals and sweat rates. Reproduced with permission⁶⁶, Copyright 2024, IEEE. b Schematic diagram of a portable photoacoustic combined therapy. Reproduced with permission⁶⁷, Copyright 2020, Elsevier. (c) integrated electrophysiological and electrochemical sensing electrodes in the ear; White dashed lines indicate their positions⁴⁰ Copyright 2024, Xu et al. d Optical image (left) and sensor array layout (right) of the sensor used to probe in-plane heat conduction. between temperature and distance at different initial temperatures. Reproduced with permission. Reproduced with permission⁶¹, Copyright 2024, IEEE.

The types of brain signal acquisition

Brain vital signs information is the basis of brain disease diagnosis. The signal is characterized by strong interference, chaos and randomness. Brain signals mainly include the intracranial pressure^68,69, neurotransmitter⁷⁰, electroencephalogram^37,45, electroocular signal⁷¹ and cerebral oxygen saturation etc. (Fig. 7).

Fig. 7. The performance verification of various flexible brain sensors in the brain.

a Response of different sensors to time-varying temperature. Reproduced with permission⁶¹, Copyright 2024, IEEE. b Response of each layer of electrostatic sensor interface to UV-vis transmission spectrum⁷¹ Copyright 2023, Shi et al. c Correlation between each indicator and ICP. Statistical parameters included ONSD, HH, Age, Size, HTN, etc. Among them, ONSD was the most closely related to ICP with SHAP value of 5.58. The closeness between HTN and ICP was 0.09, which was the lowest⁶⁸ Copyright 2024, Kim et al. d Left: M-mode image (top left) and ECG signal (bottom left) were acquired from the parasternal by wearable ultrasound. Right: The different phase of a beating cycle of the heart was shown²¹ Copyright 2023, Hu et al.

Intracranial pressure (ICP) is the pressure on the contents of the cranial cavity, including brain tissue, cerebrospinal fluid and blood. Abnormal ICP can disrupt normal blood flow to the brain, leading to symptoms such as dizziness, headaches, and vomiting. Continuous monitoring of ICP had been a challenge. Currently, monitoring ICP requires the implantation of sensors in the brain. However, due to the sensor is large, it cannot be implanted through minimally-invasive surgery, resulting in a high risk. In order to avoid the risk of craniotomy, Zang creatively proposed ultrasonic super compose gel sensor. Numerical simulation and animal experiments showed that the sensor was highly reflective of incident ultrasonic waves (Fig. 8a)⁶⁹. Implantable FBES is an important solution of ICP monitoring, which provides a valuable guidance for clinical diagnosis and treatment.

Fig. 8. The representation of brain signal acquisition.

a Implantable biodegradable materials probe intracranial pressure⁶⁹ Copyright 2024, Tang et al. b Schematic diagram of the electrostatic interface based on TENG Analysis of the interface array structure⁷¹ Copyright 2024, Shi et al. (c, d): (c) 3 d printed circuit board of wearable EEG amplifier, (d) Experimental model and visualization of electroencephalogram and electroocular signal. Reproduced with permission⁷⁶, Copyright 2024, IEEE.

The neurotransmitters are chemicals that carry messages between neurons in the brain, such as acetylcholine, monoamines and neuropeptides. The neurotransmitters are diagnostic markers of conditions, such as depression, anxiety, mental disorders and schizophrenia etc.⁷⁰. However, the neurotransmitters are based on chemical information, and the character enables the FBES to increase the misjudgment probability during acquisition. Chen et al. reported a chemically-mediated artificial neuron, where neurotransmitter dopamine could be received and released and monitored by artificial neurons monitor by such a carbon electrochemical sensor⁷². The neuro-probes are important tools for monitoring neurotransmitters, and reducing the size of implantable probes can prevent displacement during continuous monitoring and coincidently lower the risk of inflammation after the surgery. At the university of California, Anne m. Andrews’s team developed an implantable receptor field effect transistor neural probe. The field effect transistor neural probe enabled the detection limit of 5-HT to reach femtomolar with low bioflavage⁷³. In addition, researchers also focused on the biocompatibility of neurotransmitter chemical monitoring and probe stability.

The electroencephalogram (EEG) is a common brain signal. The EEG signals have a low intensity, typically on the order of microvolts(μV) and a frequency range below 50 Hz. The EEG signals were first described in 1929 and applied to BCI technology in 1973^74,75. Hu developed an integrated and high signal-to-noise ratio hybrid BCI based on wearable EEG amplifier (Fig. 8c, d)⁷⁶. To avoid signal attenuation and distortion caused by skull defects, Li proposed a hybrid BCI system that involved EEG and magnetoencephalography. Neuroimaging technique is not affected by volume conduction and have been shown to deliver BCI performance⁷⁷. However, BCI systems relying on EEG still encountered many challenges, such as optimizing signal processing methods, expanding functionalities, and assessing reliability⁷⁸. In addition, cochlear nerve action potential monitoring after vestibular schwannoma resection surgery can enhance surgical outcomes. For example, Zhou et al. introduced stretchable PEDOT:PSS guideway electrodes for the stable and continuous monitoring of CNAP values during simulated surgery⁷⁹. However, there are still some problems for EEG signal acquisition. Typically, the test results using BCI system based on EEG at different time or different physical conditions are different even though the same test parameters are adopted, and the EEG signal also lacks accuracy.

The electroocular signal (EOG) has high frequency and strong intensity, which is easily captured by FBES. The occipital EEG alpha inhibiting wave can be harnessed to detect eye closure or opening, determining that it is a good way to monitor EOG⁸⁰. Eye movement tracking provided a tool and basis for analyzing visual attention and thinking. Shi had proposed an electrostatic sensing interface, and it was used in an active eye tracking system that relied on electrostatic effects. The electrostatic charge density of the interface reached 1671.1 μC/m² after 1000 cycles, which enabled precise EOG detection with an angular resolution of 5 degrees⁷¹. With the increasing number of rare diseases and the elevating requirement of diagnosis accuracy, single brain signal detection no longer meets the clinical needs. Therefore, it is an effective solution to improve the performance of a hybrid BCI system by combining multiple important signals⁸¹, such as brain oxygen saturation⁵⁹, photoelectric fusion⁸², etc.

The sensing mechanism and characteristics of FBES

In previous studies, solid-state sensors were utilized to measure brain signals. Metal electrode sensors and Ag sensors have been widely used. However, the rigid sensors with low skin coupling led to stronger interference, necessitating the use of electrodes or conductive paste as a medium. In recent years, FBES have emerged as crucial components in various applications such as skin interfaces⁸³, wearable electronic devices and brain-computer interfaces^84–86. The operational principles of FBES encompass resistive, capacitive, and piezoelectric functionalities.

The FBES based on piezoresistive sensors

The FBES based on piezoresistive sensors aims to detect signals based on the change rate of conductive property inside the semiconductor material when they are exposed to pressure. When subjected to an external force, the charge carriers n and mobilityμinside the conductor will vary. The measurement of the piezoresistive effect is described by the piezoresistive coefficient $\mathrm{\pi }$ in Eq. (1):

$$\mathrm{\pi }=\frac{\mathrm{\Delta }\rho }{\sigma \rho }$$ 1

Where,$\rho$ is semiconductor resistivity, $\mathrm{\Delta }\rho$ is the change of resistivity, $\sigma ={\rm E}\epsilon$ is stress, ${\rm E}$ is the elastic modulus of the material, $\epsilon$ represents the strain. However, most flexible material piezoresistive coefficients have no linearity, the ratio of the amount of pressure change to the amount of electrical signal change is measured by sensitivity S, described in Eq. (2):

$$S=\frac{\mathrm{\Delta }R}{R_0{\mathrm{\Delta }}_P}=\frac{\left(R_t-R_0\right)}{R_0{\mathrm{\Delta }}_x}$$ 2

Here, ${\mathcal{R}}_0$ is the initial sensing resistance, and ${\mathcal{R}}_t$ is the value of the resistance at time t. $\mathrm{\Delta }\mathcal{P}$ is the pressure sensor value. The sensitivity of FBES is directly proportional to its performance. The piezoresistive transmission mechanism consists of band structure change, seepage theory, tunnel effect and interfacial contact resistance change⁸⁷. Band structure change describe the phenomenon where external stimuli induce alterations in semiconductor energy bands and further accordingly bring about mobility change, ultimately causing macroscopic resistivity variations in semiconductor materials (Fig. 9b)⁸⁸. Ren Yuan et al. utilized the single-electron approximation method to establish the PD-Ge band structure model⁸⁹.

Fig. 9. The characterization and properties of flexible piezoresistive materials.

a Hybrid process diagram of mGN material⁹⁰ Copyright 2019, Wang et al. (b) Energy diagram of PD-Ge conduction band energy valley level⁸⁹ Copyright 2017, Yang et al. c Schematic diagram of a circuit through pressure expansion⁸⁸ Copyright 2017, Haniff et al. d Numerical simulation of ΔI/I0 pressure on mGNEcoFlex composites. The illustration is a magnification of ΔI/I0 and pressure⁹⁰ Copyright 2019, Wang et al. e Static test results of contact area and pressure of indium‑tin-oxide-coated glass and gold-coated rubber. Reproduced with permission⁹², Copyright 2019, Elsevier.

The percolation theory states that the conductive properties of composite materials are related to the volume of conductive phase filler. When the volume of conducting phase is below the percolation concentration, the composite is an insulator. The uniform distribution of conductive phase volume shapes conductive network with the strongest conductive performance, and then it recedes. Thus, the volume flow rate is the core of conductive phase (Fig. 9a)⁹⁰. The tunneling effect refers to the process by which electrons with low energy levels cross a higher energy barrier and generate a current. The tunneling resistance of two near adjacent conducting phases can be estimated and described by the following Eq. (3)⁹¹:

$${\mathcal{R}}_s=\frac{V}{AJ}=\frac{h^{2}d}{A{e}^2\sqrt{2m\lambda }}\exp \left(\frac{4\pi d}{h}\sqrt{2m\lambda }\right)$$ 3

Where, $V$ is a potential difference, $A$ said current path to the cross-sectional area, $J$ is the current density, $h$ is the Planck’s constant and $h$ = 6.62607015×10^-34J·s. $d$ is the distance between two adjacent conducting phases, e is the unit electron charge, $m$ is the single electron mass, and $\lambda$ represents the height of the energy barrier.

Seepage theory and common tunnel flow effects on the composite conductive network were explored in this study. In order to enhance sensor sensitivity, researchers developed a design pattern centering on the variation in interface contact resistance (Fig. 9e)^92,93. Such sensors can be categorized into two types, one is the interface between electrode and the contact resistance of active layer materials⁹⁴; and another is the contact resistance within the active layer materials themselves⁹⁵. In the future, FBES based on piezoresistive sensors are expected to advance towards integration, functional diversification, low power consumption, and microstructural refinement.

The FBES based on capacitive sensors

Flexible capacitive electronic sensors began in 2000⁹⁶, and were made by composite material consisting of a dielectric layer and flexible capacitor electrode layer, described in Eq. (4):

$$\mathrm{C}=\frac{{\epsilon }_0{\epsilon }_{r}s}{d}$$ 4

Here, ${\epsilon }_0$ is the capacitive vacuum permittivity, ${\epsilon }_r$ is the relative permittivity of the composite dielectric layer, $s$ is the area of the electrode plate, and $d$ is the distance between the two flexible electrodes. The structure as shown in Fig. 10.

Fig. 10. Flexible brain capacitance electronic sensor structure.

a No external effect, (b) The state of being subjected to external pressure, (c) The state of being stretched or compressed.

In addition to the traditional flat-plate electrode plate, there were also cross-comb structure capacitive sensors that have been developed in recent years⁹⁷. It was described in Eq. (5):

$$\mathrm{C}=(\mathrm{\eta }-1)\frac{{\epsilon }_0{\epsilon }_r\mathit{wm}}{d}$$ 5

${\epsilon }_0$ is the capacitive vacuum dielectric constant, ${\epsilon }_r$ represent the relative dielectric constant of the composite dielectric layer, $m$ is the area of the electrode plate, $d$ is the distance between the two flexible comb electrode plates, $w$ is the overlap length of the two electrodes, and $\mathrm{\eta }$ is the number of electrodes. Flexible capacitive sensors are required to resist material deformation in the process of acquiring vital signs information, so it should have better fatigue resistance⁹⁸. However, the sensitivity and response time of these flexible capacitive sensors are slow. Bao et al. conducted microstructure treatment on the dielectric layer and found that a pyramid-shaped surface microstructure was able to increase the sensitivity of capacitive FBES by about 30 times (Fig. 11a)⁹⁹. In 2024, Moon et al. fabricated a PDMS-based ECoG electrode array and demonstrated that it was reliable for long-term implantation in vivo¹⁰⁰. Flexible capacitive sensors also overcome the disadvantages of fragmented output signals of resistive sensors¹⁰¹, and has a broad application prospect in the domains of arterial flow monitoring and ICP detection⁶⁹.

Fig. 11. Sensing mechanism and characteristic testing of FBES.

a Pressure response of the microstructured PDMS⁹⁹ Copyright 2010, Trung et al. b Left: Placement of the ultrasound patch and schematic diagram of the structure of cerebral arteries, right: diagram of the layered structure of the ultrasonic patch¹⁰⁴ Copyright 2024, Zhou et al. c Schematic diagram of SB-TENG rolling soft spheres with and without PTFE powder on Cu surface; (d) Open circuit voltage and transfer charge using SB-TENG with and without two soft spheres with PTFE powder. Reproduced with permission¹⁰⁵, Copyright 2023 Wiley‐VCH GmbH.

The FBES based on piezoelectric principle

Piezoelectric effect specifically refers to the phenomenon that when the conductive medium is deformed by external force in a certain direction, its internal polarization phenomenon will occur, and at the same time, positive and negative opposite charges appear on its two opposite surfaces. When the sexternal force disappears, the conductive medium will return to the uncharged state. Piezoelectric sensors generate charge by crystal polarizing under mechanical stress, and this energy conversion process can function in the absence of an external power supply. However, piezoresistive effect is produced by altering carrier mobility and band structure of the material to result in a change in resistance when subjected to stress. Regarding this, the generation of a signal in piezoresistive sensors necessitates an external power supply. In the piezoelectric FBES based on piezoelectric effect, the polarized charges inside the conductor moves in the opposite direction when subjected to an external action, and in other words, positive and negative charges migrate on the surface of opposite polarity, which induces potential difference. The piezoelectric FBES has fast response speed and high sensitivity, which is suitable for continuous signal monitoring. The sensitivity parameter ($\mathrm{s}$) of FBES based on piezoelectric principle is related to the potential difference of positive and negative charge on the resilience, and the specific description is in Eq. (6):

$$\mathrm{s}=\frac{L_0\mathrm{\Delta }\mathrm{V}}{\mathrm{\Delta }\mathrm{L}}$$ 6

$\mathrm{\Delta }\mathrm{V}$ is the change of output voltage, $\mathrm{\Delta }\mathrm{L}$ is the length of the deformation sensor, $L_0$ is the natural length of sensors.

Mochizuki et al. used AIN material made of flexible piezoelectric sensor to attach to skin surface and identify facial muscle movement¹⁰². However, the stretchability of flexible sensors made of inorganic materials is weak, while the flexible sensor made of organic material encounters unstable structure. In order to overcome the shortcomings of two types of sensors, Chen reported an FBES based on nanofibers & ZnO nanowires composite material. The FBES had a tensile ratio up to 30% and a bending angle up to 150¹⁰³. In practical application, most flexible piezoelectric pressure sensors are difficult to achieve static pressure measurement. Especially in the measurement of blood flow spectrum of cerebral arteries, the special structure of skull renders ultrasound significantly attenuated, making the detection of intracranial blood vessels become a technical bottleneck (Fig. 11b)¹⁰⁴.

The multi-mode fusion with FBES is a hot topic. It combines the advantages of various kinds of sensors and offers a potential strategy to enhance sensor performance. A recent study introduced a capacitive and pressure dual-mode sensors based on a complementary spiral design and dielectric calcium cross-linked alginate film. Theoretical analysis and numerical simulations showed that the electrode pattern design with a higher specific stripe length significantly improved the intensity of the stripe field¹⁰¹. In addition, with the progress of the synthesis of materials and the semiconductor technology, flexible field effect transistor sensors have also been used in wearable BCI, such as Triboelectric type strain sensor (Fig. 11c, d)¹⁰⁵. In the future, self-powered flexible sensors are promising in the field of wearable brain-machine interface. The effect of sensing mechanism on the performance of FBES is described in Table 1.

Table 1.

The Effect of sensing mechanism on the performance of FBES

Transmission mechanism	Materials	sensitivity	Electric conductivity	Response time/ms	Mechanical property	Reference
Piezoresistive	EcoFlex+Oxide/mGN	1302.1 kPa−1	0.003 S·m−1		122.8 kPa	90
Piezoresistive	AgNWs	1.5 kPa			Percolation threshold was 0.48 vol%	94
Piezoresistive	PDMS/MWCNTs	0.183%/°C			Good stability at 30–50 °C	112
Piezoresistive	MXene aerogel	1,799.5 kPa−1		11 ms	Cycles: å 25,000	119
Capacitive	Ionic gel	72.86 kPa−1		43 ms	Cycles:7300	118
Piezoelectric	Hydrogel+Ag		374 S/cm		Compressive ability: 10 Kpa	114
Piezoelectric	PVDF@ZnO	G = 4. 59			Curvature:150°	103
Hybrid sensing	PDMS + CB/CNTs	G = 4.36			10%	111
Capacitive+Piezoresistive	Dielectric calcium cross-linked alginate film			10 ms		101

The materials and structures of FBES

Flexible substrate materials, conductive sensor and flexible electrode are indispensable components of FBES. To enhance the performance of FBES for optimizing clinical diagnosis and treatment strategies, researchers focus on selecting suitable materials and optimizing their internal structures.

The choice criterion of materials

FBES that is attached to the surface of brain skin needs to meet the fatigue resistance and stretchability. In the past, researchers tested the conductivity, softness and sensitivity of materials through experiments or theoretical calculations. The polyethylene glycol terephthalate shows a high flexibility and is identified as an ideal material for anti-fatigue substrate^106,107. Polyimide (PI) is a kind of polymer that can withstand high temperature and extreme low temperature, which has been widely used as a substrate for FBES. Sahay et al. compared the properties of nickel-niobide oxide nanolayers with pure nickel nanolayers on a PI substrate, and found nickel-niobide oxide nanolayers a potential material for electrodes in FBES (Fig. 12a)¹⁰⁸. The new diamine doping in PI could accelerate the motion of polyimide molecular chains and improve the processability of PI that has been proved to be a good biocompatible flexible material. Multilayer polyimide probes were implanted into the parietal cortex of mice, and the stability of neuronal signals was recorded at 1, 4 and 6 months, including the rate stability of electrode firing, pulse count change, and the amplitude and duration of signals. After 180 days of recording, the micro-movement of flexible probes was very small, and the parietal cortex of mice was also unaffected. Scanning electron microscope analysis showed no signs of delamination or degradation in the electrode material¹⁰⁹. Polydimethylsiloxane (PDMS) is not a conductive material, but ideal conductivity can be obtained by doping filler¹¹⁰. PDMS has a good coupling with the skin, and it is an ideal substrate for FBES. Based on the idea that the mixture of conductive materials and PDMS could improve substrate performance, Zheng et al. prepared two types of FBES by mixing PDMS with carbon black and one-dimensional carbon nanotubes, respectively¹¹¹. Yang et al. mixed the conductive material of MWCNTs with PDMS as filler and fabricated PDMS/MWCNTs composite with negative temperature resistance coefficient¹¹². The flexible surface electrode made of PDMS was attached to the skin to monitor electrical signals. The test showed that the electrode maintained high quality recording ability (comparable to commercial electrode quality) and the signal-to-noise ratio remained stable for 3 weeks. These results provided evidences that the flexible electrode was compatible with biological tissues¹¹³. In addition, copolyester natural rubber and thermoplastic polyurethane have been used as substrate materials for FBES. The synthesis of composite materials is an important strategy to expand the performance of FBES.

Fig. 12. The different types of FBES material selection and structure.

a Left and right: TEM cross section micrographs of Ni-Nb2O5-PI, left is 200 nm and right is 100 nm¹⁰⁸ Copyright 2023, Sahay et al. b Schematic diagram of the ionic electron bilayer hydrogel device structure, which consists of a styrene-ethylene-butene-styrene substrate, an electronically conducting hydrogel, a stretchable silver-based interconnect, an elastic encapsulant, and an ionically conducting hydrogel matrix²⁵ Copyright 2024, Arwani et al. c Schematic diagram of MXene aerogel for self-healing flexible electronic skin and pressure sensing. Reproduced with permission¹¹⁹, Copyright 2023, American Chemical Society. d Conical light sensor array micropillar structure; e Illustration of color blind image perception by artificial tip cone photoreceptor array¹²⁵ Copyright 2024, Wang et al.

The flexible electrode plays a crucial role in signal acquisition, and has important effects on the electric transformation, such as gallium indium alloys, and conductive hydrogels. Majidi et al. mixed silver with a polyacrylamide-alginate brine gel to create flexible electrodes, and the conductivity reached 374 S/cm¹¹⁴. The poor accessibility of hydrogels to weak biofluids limited the continuous monitoring application of these electrodes in detecting brain signals. Liu et al. studied an ion electron double water gels to promote the sequence dissolution, diffusion and electrochemical reaction of solid analysis, where the continuous monitoring performance of the mixture was verified (Fig. 12b)²⁵. The ideal tensile properties and good fatigue resistance of materials are beneficial for the stability of FBES signals. In order to obtain stable mechanical properties of flexible semiconductor devices, researchers used chemical and physical methods to improve performance¹¹⁵. However, high flexibility and conductivity are rarely achieved simultaneously under external forces. Recently, Bao’s team reported PSC films with thicknesses less than <100 nm, and the used polymer has a carrier mobility of 0.2 cm²/(V·s) under 100% biaxial strain, which makes it an ideal substrate material for fabricating FBES¹¹⁶. Electrode materials are the core of FBES, and their structural innovation and material selection innovation are still the research focus. Hydrogels and their composites are ideal electrode materials. We cited typical and common flexible substrate materials in Table 2 with an emphasis on the properties and composition.

Table 2.

Performance comparison of flexible substrate materials

Substrate materials	Polyethylene glycol terephthalate	Polyimide	Polydimethylsiloxane
Elastic modulus	0-120° bending could be achieved without affecting performance.		The electrode had a Young’s modulus of 0.0545 MPa
Biocompatibility		the probe micromovement were minor and it recorded neuron for over 6 months	The electrodes maintained their ability to record signal over 3 weeks.
Electrical conductivity	The change in resistance ratio ΔR/R0 was 0.18, t = 75 s.	The conductivity is 3.6 kΩ/cm	The promising conductivity of 9.62 × 10−3 S/cm
Dimension	Thickness was 25 μm	Thickness was 100 nm	1 cm diameter and 1 mm thickness
Reference	107	109	113

Structure optimization

The sensitivity, response time, fatigue resistance and stability of brain signals are key criteria to evaluate the performance of FBES. Therefore, the structural design of FBES not only affects the acquisition and transmission performance, but also is an effective strategy to improve the performance of flexible sensors, which generally involves the designs of flexible base and conductive sensor material microstructure¹¹⁷.

A hierarchical structure is prevalent in FBES owing to its clear principles and suitability for electronic skin, and its design principle is clear and suitable for the preparation of electronic skin¹¹⁸. However, the spacing and self-packing tendency of inter-layer structure hamper the alteration of electron channels under external pressure. To solve the issue, Li proposed a rapid gas foaming strategy to develop an inter-layer adjustable MXene aerogel. The interlayer network structure of MXene generated sufficiently large electron channels under pressure, which adequately utilized the metal characteristics of MXene (Fig. 12c)¹¹⁹. Layered structure has become a common design idea for FBES, which actually has been widely used in electrochemical sensors and biosensors⁴⁰.

Microstructure optimization of electrodes can improve the sensitivity of flexible sensors, and here microstructure design covers fold, porous, etc¹²⁰. In order to improve the performance of micro-structured flexible sensors to adapt to wearable physical sign monitoring, Ma et al. introduced a high-sensitivity pressure FBES. The flexible substrate coated with AgNWs was used as the top/bottom electrode material, and the microarray structure ensured the high sensitivity of FBES¹²¹. Recent studies have showed that the optimized microstructures was not only beneficial for flexible capacitance sensors, but also could be applied to flexible resistive and piezoelectric sensors^122,123.

The concept of bionic structures highlights mimicking the structure and characteristics of natural organisms to enhance performance. In order to improve the stability of wearable BCI, an interwoven neural network was formed at the interface through mesh flexible electronic components in the past. For example, biomimetic designs have been used to fabricate flexible electronic devices to mimic the structural and mechanical properties of single neurons. Sun et al. constructed a 1024 pixels flexible light sensor wherein a composite of carbon nanotubes and perovskite quantum dots as a signal acquisition material was used for neuromorphic vision systems¹²⁴. The cones inside eyes can convert natural light into neural signals, and then these signals were reflected in brain to recognize color (Fig. 12d-e)¹²⁵. In addition, the matrix material has been prepared as sponge foam or carbon skeleton¹²⁶, and the serpentine structure, spiral structure and porous structure were also widely used after rational engineering. The FBES with different structures showed different properties, e.g., DNA models could be bent up to 90° ¹²⁷. Different FBES are described in Table 3, wherein their structure, composition and performance are elucidated and discussed.

Table 3.

The performance of different structure FBES

Structure	Materials	Performance	Reference
Compound eye microstructure	PDMS+Ag	The sensitivity of the FBES reached 0.32 kPa-1; the response time≤130 ms, and recovery time≤120 ms; the hysteresis parameter is less than 7%, which could withstand 12,000 impacts	100
Crater microstructure	Dielectric calcium cross-linked alginate film	The detection distance was about 55 mm, the response time was 10 ms, and the linear output R2 of the response was 0.995.	11
Hierarchical structure	EDL & ionic gel	Sensors showed a high sensitivity of 72.86 kpa-1, the pressure sensing range was 1 Pa to 400 kPa, the response time was 43 ms, the number of cycles was 7300.	118
Hierarchical structure	Mxene	The sensitivity was 1799.5 kPa-1, the response time was 11 ms and the Young’s modulus was more than 25000 cycles.	119
Bionic structure	Carbon nanotubes & perovskite quantum dots	The sensitivity was 5.1 × 107 a /W and the detection rate was 2 × 1016.	136
Fiber structure	Hydrogel	The hydrogel could self-repair within 10 minutes after deformation, the tensile strength was greater than 7000%, and the conductivity was 11.76 S/cm	137

Conclusions and outlook

With the interdisciplinary development of materials science and electronics, FBES has made unprecedented progress. The advances in selecting and manufacturing flexible brain electronic sensors directly impact the performances of wearable BCI devices. Therefore, this paper provides a comprehensive review of recent advancements in wearable brain-computer interface technology, brain signal collection methods and types, sensing mechanisms of flexible brain electronic sensors, material selection, and structural optimization. The successful clinical application of wearable brain-computer interface devices hinges on wearable brain computer interface devices, which eventually relies on the preparation of optimal flexible brain electronic sensors. Currently, challenges faced by flexible brain electronic sensors include the shielding effect of skull that attenuates signal, the biocompatibility of artificial materials and the human body. These challenges hinder the expansion of wearable brain computer interface devices in clinical practice. Despite the fact that the accuracy rate of intelligent medical treatment based on machine learning has been improved, the accuracy of machine learning-based intelligent medical diagnosis of brain diseases fails to meet the clinical application requirements. Ultimately, the successful clinical application of wearable SCI devices is decided by ideal FBES, and ideal FBES is also essential for advancing wearable brain computer technology. To address current challenges, future research will be followed along the several directions.

Personalized diagnosis and treatment

Signal acquisition is the precondition for disease diagnosis in the field of intelligent medicine, where the integration of diagnosis with treatment is common and pivotal. Individual differences and external disturbances will interfere with biochemical signals to some degree, necessitating innovative solutions to enhance signal accuracy, which therefore demands the precise collection and processing of complex physiological data using FBES. According to the type and location of diseases, tailored professional FBES are made to establish a correlation rationale between signals and diseases, and improve the efficiency and accuracy of diagnosis. Based on the diagnosis conclusion and the adaptive feedback treatment approach from BCI system, the closed-loop interaction between diagnosis and personalized treatment is built. The advancement of monitoring, diagnosis, and treatment has extended the utilization of BCI and FBES in smart medicine, propelling the field towards the convergence of conventional and intelligent medicine.

Low power operation

Power consumption directly affects the overall device performance and user experience. Advances in FBES keep pace with the evolution of integrated circuits, showing the trends of integration and multi-functionality in the future, and the design of integrated processes can reduce the operating power of wearable BCI systems^128,129. The fusion of multi-mode FBES enables the design of miniaturized electrodes by manufacturing flexible electrodes, and integrating multi-mode sensors such as electroencephalography, near-infrared spectroscopy and electromyography into a unified platform is available for reducing the number of electrodes. In order to ensure the balance between the acquired signal quality and chip power consumption, the design of system on chip (SOC) is considered to integrate low noise amplification, filter and digital-to-analog conversion to reduce discrete devices. Ultra-low power chips like AFE chips utilized sub-threshold operation (such as Texas Instruments ADS1299). In addition, the flexible PCB substrate is combined with SOC, such as PI or PDMS substrate, and 3D packaging technology can be also used to integrate the analog front-end, SOC and communication module into the flexible PCB. Such an integration adapts to the curved surface of head, thereby reducing the power consumption and motion-related artifacts, so as to achieve the low-power operation of wearable system. The pursuit of low power consumption is key, allowing FBES to operate longer, collect data more continuously, and promote wearability. The BCI system integrates a variety of functional units, including composite materials, such as composite materials, power sources, sensors, actuators, and communication modules. In addition, self-electric flexible sensor engineering is also a good route to reduce system power consumption^51,130. In conclusion, maintaining low power consumption and employing efficient power sources are essential to ensure the function and continuous operation of FBES as well as devices including drug delivery systems, positioning FBES at the forefront in shaping the future of health.

Machine learning AIDS

The application of FBES generates a large amount of data, and it is the basis for machine learning and statistically-assisted judgment. Machine learning, by way of extensive data training, can enhance the intelligence of diagnosis model, such as CT image-assisted diagnosis, surgical navigation system, etc. Although deep learning has been utilized to appraise the surface roughness of flexible sensors, the accuracy is only 95%¹³¹. Machine learning algorithms are capable of extracting and categorizing brain signals to build an intelligent system, which is a closed-loop system. Learning models such as convolutional neural network and deep recurrent neural network have advantages in processing real-time brain signals^132,133, laying the foundation to the clinical application of wearable BCI technology. However, the challenge in terms of medical ethics, clinical validation, and the convergence of flexible electronics with intelligent medical systems remain unresolved, and addressing these constraints is of great importance for optimizing clinical operations and expanding disease-specific monitoring.

Multimodal information fusion

To improve the accuracy of clinical diagnosis and treatment with multiple indexes and directions, the transition of FBES from single signal acquisition to the integration of multi-mode signals is highly desirable. The function of FBES extends from one-way brain vital sign information collection to bidirectional communication pathways between machine and brain, brain-to-brain interactions, and brain-to-machine interfaces^134,135 (Fig. 5b). However, solving cross-sensitivity and sensing crosstalk between mixed physical and biochemical signals remains a challenge. In addition, the flexible deep brain electrode is confronted with obstacles related to biocompatibility, and the assurance of consistent operation of FBES in extreme environments poses a significant challenge. As a result, the chase for new composite flexible materials to facilitate simultaneous monitoring of diverse indicators and address biosafety issues is imperative for gaining insights into the pathogenesis of brain disorders.

Material innovation compatible with biosafety

The progress of BCI keeps pace with the development of FBES. Flexible deep brain electrode is an effective means to detect deep brain signals, but it still faces the challenges of biocompatibility and safety. Implantable FBES materials are still in the laboratory stage, lacking evidences in the clinical application. Ensuring reliable and secure operation of FBES under complex clinical conditions associated with medical ethics, individual differences and environmental complexity, is another daunting challenge⁵⁹. In addition, the hybrid flexible electron of FBES and biomedical coating can meet more clinical needs, and the biomedical coating can replicate the surface characteristics of tissues, such as roughness and hydrophilicity, etc. Concurrently with BCI technology advances, biosafety will also play a more important role in achieving the continuous integration of materials science and medical engineering.

Author contributions

J.L. and G.C. contributed equally to this work. K.Z. and J.L. proposed this theme, organized the skeleton, provided the raw materials and selected the cited references. J.L., G.L., and L.X. re-organized figures and wrote the manuscript. J.L. wrote this manuscript and K.Z. and R.J. revised the manuscript, and supervised and supported the project. All authors commented on this manuscript.

Data availability

No datasets were generated or analysed during the current study.

Competing interests

The authors declare no competing interests.

Statements and Declarations

This article does not contain any new studies with human participants or animals performed by any of the authors.

Ethical approval

This is a systematic literature review, therefore, no ethical approval is required.