Brain–Computer Interfaces: A New Era in Treating Brain and Mental Health Disorders

Table of Contents

• Key Points
• Background: Why Brain–Computer Interfaces Matter
• The History of Brain–Computer Interfaces
• How Brain–Computer Interfaces Are Classified
• Treating Movement Disorders with Brain–Computer Interfaces
• Helping Patients with Communication Barriers
• Psychiatric Conditions: Diagnosis and Treatment
• Challenges and Limitations
• Future Directions and Emerging Technologies
• What This Means for Patients
• Recommendations and Takeaways
• Frequently Asked Questions
• Source Information

Key Points

• BCI lets people control devices with their own brain signals and can provide feedback for rehabilitation and closed-loop neuromodulation.
• Non-invasive BCIs are safer and less expensive but have poorer signal quality; invasive BCIs have the highest signal-to-noise ratio but require surgery.
• BCI is being tested for movement disorders, communication loss, and psychiatric conditions such as autism and depression.
• Challenges include long-term electrode safety, limited signal resolution, portability, and the need for better personalized signal decoding.
• New electrode materials, endovascular BCI, optogenetics, and artificial intelligence integration are expected to enhance future BCI systems.

Background: Why Brain–Computer Interfaces Matter

Neurological and psychiatric diseases are increasing worldwide. Population aging, longer life expectancy, and psychological stress all drive this rise. These conditions place heavy social and economic burdens on families, healthcare systems, and economies.

The numbers are staggering. According to 2021 statistics from the Global Health Data Exchange, an estimated 2.87 billion people worldwide are affected by neurological diseases. These include age-related dementias, neurodegenerative diseases, strokes, and traumatic brain injuries. Another 1.09 billion people suffer from psychiatric conditions such as depression.

Depression alone reached 320 million cases globally in 2021. That equals 43 cases per 1,000 individuals. Parkinson's disease (PD) affects 149 individuals per 100,000 worldwide. Parkinson's disease has led to the loss of an estimated 7.47 million healthy life years.

Dementia cases from Alzheimer's disease (AD) and other causes approach 55 million globally. In the same year, there were about 12 million stroke incidents worldwide. Among stroke survivors, 53.4% experienced post-stroke cognitive impairment.

These disorders often interact with each other. The underlying neuropathological mechanisms can worsen disability. In summary, neurological and psychiatric diseases have become a major global public health crisis. This crisis demands urgent development of innovative therapies.

However, the adult human brain has limited repair capacity. Neurological damage remains largely irreparable despite decades of intensive research in regenerative therapies.

Brain–computer interface (BCI) technology is a promising research direction for treating these diseases. BCI systems provide users with brain output channels. These channels convert neural activity into control instructions for external devices. BCI can also provide input signals as motor, tactile, or visual feedback. This feedback supports rehabilitation training and enables personalized, closed-loop neuromodulation therapies.

The emergence of BCI is redefining the boundaries of modern medicine. It provides a new technical path for treating neurological and psychiatric diseases.

A typical BCI system has three key components:

• A signal acquisition module
• A signal processing module
• Controllable external devices

Currently, BCI in medicine focuses on neuropsychiatric diseases. Its uses include disease assessment, functional assistance, rehabilitation training, and neuromodulation.

For diagnosis, BCI systems can detect the onset of neurological impairments or events such as seizures. For functional assistance, BCI-generated output signals let patients with strokes, limb defects, or spinal cord injuries (SCIs) control external devices such as wheelchairs and prosthetics. For rehabilitation, BCI provides essential feedback and training paradigms. BCI can also trigger neuromodulatory inputs such as deep brain stimulation (DBS) to suppress seizures or regulate movement in Parkinson's disease.

BCI systems can also be designed to evaluate and treat psychiatric disorders. These include depression, autism, attention deficit hyperactivity disorder (ADHD), and bipolar disorder. Treatment may use stimulation-based or pharmacological interventions.

Despite this promise, several challenges remain. These include risks from implanting signal detectors and stimulators, limited system portability, and the need for better neural signal decoding for personalized device control.

This review summarizes recent advances in BCI system design for neuropsychiatric diseases. It covers signal detection and decoding, feedback control for external devices and neuromodulation, technical achievements in system engineering and fabrication, and potential future applications.

The History of Brain–Computer Interfaces

The development of BCI systems traces back to the discovery of bioelectric phenomena in the 19th century. In 1875, British physicist Richard Caton was the first to record electrical activity from animal brains. He published his findings in the British Medical Journal. This discovery paved the way for routine clinical use of electrophysiological measurements. These include the electroencephalogram (EEG), electrocardiogram, electromyogram (EMG), and compound nerve action potential recordings. It also laid the foundation for BCI technology.

In 1913, Prawdicz-Neminski was the first to capture seven distinct patterns of bioelectrical activity in animal brains. He called them components of the "electrocerebrogram." However, he misinterpreted these signals as artifacts from human interference. He did not recognize them as originating from biological processes.

In 1924, Hans Berger successfully recorded the first human EEG. He identified endogenous oscillating waveforms, now recognized as α- and β-waves. This milestone provided a crucial theoretical foundation for BCI technology.

The mid-20th century brought significant progress. In 1964, Walter et al. first measured slow cortical potentials (SCPs). These were later adapted to enable communication for patients with locked-in syndrome (LIS). LIS is a condition of full awareness but complete paralysis, also known as pseudocoma.

In 1969, Eberhard Fetz conducted the first BCI experiments with monkeys. This marked a pivotal breakthrough. Around the same period, M.B. Sterman identified the sensorimotor rhythm (μ-rhythm) in feline brains.

Despite these advances, the term "brain–computer interface" was not formally introduced until 1973. Jacques J. Vidal of UCLA developed a system based on visual event-related potentials (ERPs). This established BCI as a distinct research field.

A landmark achievement followed in 1978. William Dobelle implanted a 68-electrode array in a blind patient's visual cortex. Electrical stimulation successfully induced phosphenes, which are light perceptions without actual light input. This represented the first major human application of BCI technology.

These achievements were followed by breakthroughs in BCI signal detection and processing. In 1988, Farwell and Donchin proposed the famous "P300 speller" paradigm. This allowed individuals with severe motor impairments to communicate. They identified selected letters on a screen through the P300 ERP waveform associated with decision-making.

In the early 1990s, Pfurtscheller and colleagues expanded BCI application. They used event-related desynchronization (ERD) and event-related synchronization as BCI control signals. In 1992, Erich E. Sutter developed a BCI system based on visual evoked potentials (VEPs). In 1996, McMillan et al. introduced a system using steady-state visual evoked potentials (SSVEPs).

Further advancements were made in applying BCI to control motor devices. In 1998, Philip Kennedy and Roy Bakay of Emory University implanted a "motor neuroprosthesis" in a patient with LIS. This enabled control of a computer cursor using BCI. In 1999, the first BCI International Conference was held. This marked a key step toward systematic classification and standardization of BCI technology. In 2000, Jonathan Wolpaw proposed a comprehensive definition of BCI, laying a theoretical foundation for subsequent research.

BCI system development accelerated in the early 21st century. In 2004, Matt Nagle was the first patient to use an invasive BCI system to control a robotic arm. This marked a major breakthrough in clinical application of BCI technology. Since then, experimental and clinical applications have expanded rapidly. Over the past decade, numerous BCI-based applications for humans have been developed and tested. These have attracted increasing attention from academia and news media.

How Brain–Computer Interfaces Are Classified

The general goal of BCI systems is to enable direct interactions between the human brain and external devices. This can be achieved in several ways. Neural signals can be converted into specific output commands, such as for motor prosthesis control. External stimuli can be converted into brain signals for artificial sensory perception. Reciprocal communication can also be facilitated, as seen in sensorimotor feedback systems.

BCI systems can be categorized based on signal measurement modality, application, and invasiveness. Each category has distinct advantages and limitations in signal acquisition and processing.

According to the signal source, BCIs can be divided into SSVEP-based, P300-based, and motor imagery (MI)-based types.

• Steady-state visual evoked potentials (SSVEPs) are brain waves produced in synchrony with visual stimuli. A user can generate control signals at a specific frequency in response to visual targets presented at that frequency.
• The P300 potential is a transient ERP produced when the brain receives an unexpected stimulus.
• Motor imagery (MI) refers to electrical signals generated when imagining the preparation or performance of a movement.

These signals can be further divided into evoked and spontaneous signals. Evoked activity refers to neural electrical potentials generated in the brain upon external stimuli, such as SSVEPs and P300 waveforms. Spontaneous activity refers to neural activity generated by the user, such as MI.

Other BCI signal paradigms are also worth exploring. These include error-related negativity (ERN), movement-related cortical potential (MRCP), SCP, and ERD. These paradigms have found particular utility in specific applications.

• The ERN reflects a neural mechanism for performance monitoring and error correction. Larger ERN amplitudes correlate with enhanced error detection and compensatory behaviors.
• The ERD involves attenuation of EEG power in the alpha and beta frequency bands before movement. It is widely applied in motor imagery and execution tasks.
• The MRCP exhibits distinct spatiotemporal patterns during movement preparation and execution. Studies have combined MRCP and ERD to elucidate the temporal dynamics of motor control.
• The SCP has been investigated for its role in SCP modulation, providing a low-frequency alternative for BCI control.

Depending on the location of electrode implantation, BCIs can be divided into invasive, semi-invasive, and non-invasive types.

Electrodes for noninvasive BCI are placed outside the skin. Examples include conventional EEG recording and magnetoencephalography (MEG). Electroencephalographic electrodes record electrical activity generated by vertical currents on the scalp surface. MEG measures the magnetic field generated by a tangential current.

A semi-invasive BCI is an intermediate technology between non-invasive and fully invasive approaches. Its electrodes are implanted inside the skull but on the surface of the cerebral cortex (epidurally or subdurally). They do not penetrate the brain parenchyma. Electrodes can be surgically placed on the cortical surface, such as via electrocorticography (ECoG). They can also be minimally injected as flexible components, such as hydrogel or mesh-like structures, into target brain regions.

This approach combines advantages of non-invasive methods (lower surgical risk) and invasive methods (higher signal quality). The absence of parenchymal penetration provides significantly superior spatial resolution compared to EEG. It also poses lower immunological response risks than fully implanted electrodes.

In the most invasive configuration, multi-unit EEG electrode arrays are implanted on or beneath the cortical surface. These capture neuronal signals generated near the electrode tips. ECoG electrodes can be placed on the cerebral cortex surface to record synchronized neuronal activity. They can also be implanted into gray matter, depending on the application scenario.

Non-invasive BCIs such as EEG-based systems are the safest, most secure, and least expensive. They require no surgical implantation. However, the signal-to-noise ratio (SNR) and spatial resolution are comparatively poor. Signals can be easily distorted by external signals and movement.

Invasive BCI systems achieve the highest SNR. However, they require implantation surgery and may trigger an inflammatory response at the implantation site. Semi-invasive BCI strikes a balance between the two.

Non-invasive electrodes can be further divided into dry, wet, and semi-dry electrodes. This depends on whether conductive pastes are required.

• Dry electrodes do not require conductive pastes. However, higher input impedance at the skin surface results in relatively poor signal quality.
• Wet electrodes use conductive pastes to reduce skin impedance, improve signal quality, and reduce motion artifacts. However, their adhesion may require removal of underlying hair and cuticle, resulting in skin damage. Conductive pastes can also dry out from evaporation, decreasing conductivity. Wet electrodes are not appropriate for long-term signal recording.
• Semi-dry electrodes combine the ease of use of dry electrodes with the higher signal quality of wet electrodes. They are typically made of flexible polymer materials or nano-metal porous materials. They contain electrolytes that are automatically released and penetrate the stratum corneum (outermost epidermal layer) during use. This reduces impedance and improves signal quality.

BCI systems can be classified as open loop or closed loop based on feedback mechanisms.

Open-loop BCIs analyze brain states in real-time to control external devices. However, they lack bidirectional signal flow, limiting adaptive adjustment. For example, the P300 speller paradigm enables communication by detecting ERPs in response to visual stimuli. However, it does not integrate user feedback to optimize performance. This system is suitable for static tasks like letter selection. It may struggle with dynamic adjustments.

In contrast, closed-loop BCIs integrate real-time feedback to modulate stimulation parameters or decoding strategies. These systems typically use neurofeedback (NF) to analyze brain activity. They then initiate neuroplasticity or modulate brain activity through brain stimulation techniques.

For instance, adaptive deep brain stimulation (aDBS) systems dynamically adjust stimulation intensity. They do this based on subthalamic nucleus local field potentials (STN-LFPs) in Parkinson's disease patients. This reduces motor fluctuations and improves therapy precision.

Another BCI system based on EEG incorporates adaptive closed-loop control algorithms. These automatically adjust feedback patterns and control strategies according to individual differences and training progress. This achieves more effective rehabilitation training.

Closed-loop systems demonstrate higher decoding accuracy and personalized intervention. However, they require more complex hardware and risk feedback-induced neural adaptation.

Treating Movement Disorders with Brain–Computer Interfaces

Movement disorders are the primary manifestations of many neurological diseases. Patients with Parkinson's disease, for instance, suffer from progressively worsening motor symptoms. These include tremors, stiffness, and gait disturbances.

Medications can relieve these symptoms in the early stages. However, they usually lose efficacy with disease progression. BCI systems can be designed to monitor activity in the motor cortex. They decode these signals with the assistance of various algorithms, including machine learning. They then send decoded commands to control external devices such as robotic arms and stimulators.

Since Matt Nagle became the first patient to use an invasive BCI system to control a robotic arm in 2004, BCI use for movement disorders has shown increasing promise. In recent decades, many new technologies have been developed for the brains or limbs of patients. This section describes BCI systems for movement disorders according to the technologies applied rather than specific diseases. This is because treatment methods often overlap.

Prosthetics and Robotic Arms

Spinal cord injuries (SCIs) can result in life-altering paraplegia or quadriplegia. For patients with these symptoms, BCI systems using implanted electrodes can record neural activity from the motor cortex. They decode this activity into control signals interpretable by prosthetic limbs and other assistive devices.

In addition to unidirectional open-loop systems for external device control, a bidirectional BCI system has been described for an SCI-induced paraplegia male patient. This system not only decodes neural activity from the motor cortex to control a robotic arm. It also provides tactile feedback by delivering intracortical microstimulation to the sensory cortex for enhanced control.

In another application, a subcortical stroke patient implanted with the BCI system was able to decode EEG signals associated with imagined hand movements. The patient then controlled an exoskeleton for opening and closing actions.

Further advancements have been made in decoding brain signals to control prosthetic fingers. A hybrid system integrates motor imagery (MI) with other signal detection modalities. By integrating these modalities, the hybrid system effectively mitigates the inherent limitations of each modality. For example, it addresses the low temporal resolution of functional near-infrared spectroscopy (fNIRS). This significantly enhances the precision and functionality of the prosthetic device. This advancement offers patients a more natural and flexible movement experience.

Additionally, the development of a portable and modular BCI software platform has created new possibilities for home-based applications. For instance, this platform enables control of a mechanical glove for grasping as an output device via a smartphone application. This innovation simplifies the operation of BCI systems. It also offers a more accessible rehabilitation tool for paralyzed patients, significantly improving functional independence.

Helping Patients with Communication Barriers

BCI technology offers alternative communication methods for patients who cannot speak or move. This includes patients with locked-in syndrome (LIS), a condition of full awareness but complete paralysis. The P300 speller paradigm, developed in 1988, was an early example. It allowed individuals with severe motor impairments to communicate by identifying selected letters on a screen.

BCI systems can also decode speech-related neural signals. This enables communication through brain activity rather than physical movement. These applications represent a significant advance for patients with communication barriers.

Psychiatric Conditions: Diagnosis and Treatment

BCI shows growing potential in both diagnosis and treatment of psychiatric conditions. This is particularly true for conditions like autism and depression. BCI systems can be designed for evaluation and treatment through stimulation-based or pharmacological interventions.

These systems can also be applied to other psychiatric disorders. These include attention deficit hyperactivity disorder (ADHD) and bipolar disorder. The ability to monitor brain activity in real-time offers new possibilities for personalized treatment.

Challenges and Limitations

Despite significant progress, BCI faces several challenges. The long-term biocompatibility of electrodes remains to be improved. This means electrodes must be safe for long-term use inside the body without causing harm. The resolution of neural signals also remains to be improved.

Other challenges include risks associated with signal detector and stimulator implantation. Limited system portability is another issue. There is also a need for improved neural signal decoding for personalized device control.

Invasive BCI systems achieve the highest signal-to-noise ratio. However, they require implantation surgery and may trigger an inflammatory response at the implantation site. Non-invasive BCIs are safer and less expensive. However, their signal-to-noise ratio and spatial resolution are comparatively poor. Signals can be easily distorted by external signals and movement.

Closed-loop systems demonstrate higher decoding accuracy and personalized intervention. However, they require more complex hardware. They also risk feedback-induced neural adaptation.

Future Directions and Emerging Technologies

To address current limitations, research on new electrode materials is ongoing. These include carbon nanomaterials and composites. Emerging BCI technologies present new possibilities. These include endovascular BCI and optogenetics BCI.

The integration of multimodal technologies and artificial intelligence in BCI systems is expected to enhance performance. This integration should also enable more personalized treatment.

Overall, BCI technology holds great promise for improving the quality of life of patients with neuropsychiatric disorders. It also drives innovation in the medical and neuroscience fields.

What This Means for Patients

For patients with movement disorders such as Parkinson's disease, BCI offers a potential path to regain motor function. This includes control of prosthetic limbs, robotic arms, and exoskeletons. It also includes brain stimulation-based therapies that can reduce motor fluctuations.

For patients with communication barriers, BCI enables alternative communication methods. This includes speech-related neural signal decoding. This could be life-changing for patients with locked-in syndrome or severe motor impairments.

For patients with psychiatric conditions such as autism and depression, BCI shows growing potential in both diagnosis and treatment. This represents a new frontier in mental health care.

The numbers speak to the scale of need. With 2.87 billion people affected by neurological diseases and 1.09 billion by psychiatric conditions, the development of effective BCI treatments could have enormous impact. Depression alone reached 320 million cases in 2021. Parkinson's disease affects 149 per 100,000 people worldwide. Dementia cases approach 55 million. Stroke incidents reached about 12 million in one year, with 53.4% of survivors experiencing cognitive impairment.

Recommendations and Takeaways

Based on this review, several key takeaways emerge for patients and caregivers:

1. BCI is an emerging field with real clinical progress. Since the first human EEG in 1924 and the first BCI-controlled robotic arm in 2004, the technology has advanced rapidly.
2. Different BCI types suit different needs. Non-invasive BCIs are safer but have lower signal quality. Invasive BCIs have the best signal quality but require surgery. Semi-invasive BCIs balance these trade-offs.
3. Movement disorders are a leading application. Patients with spinal cord injuries, strokes, and Parkinson's disease have used BCI to control prosthetics, exoskeletons, and robotic arms.
4. Communication barriers can be addressed. BCI enables alternative communication for patients who cannot speak or move.
5. Psychiatric applications are growing. BCI shows potential for diagnosing and treating autism, depression, ADHD, and bipolar disorder.
6. Challenges remain. Long-term electrode safety, signal resolution, portability, and personalized decoding all need improvement.
7. New technologies are on the horizon. Carbon nanomaterial electrodes, endovascular BCI, optogenetics, and artificial intelligence integration are expected to enhance future BCI systems.

Patients interested in BCI should discuss options with their neurologist or psychiatrist. Clinical availability varies by condition and region. Most BCI applications remain in experimental or early clinical stages.

Frequently Asked Questions

What is a brain–computer interface, or BCI?

A BCI is a system that lets a person control devices using their own brain signals. It converts neural activity into control instructions for external devices, and can also send signals back as motor, tactile, or visual feedback. A typical system has three parts: a signal acquisition module, a signal processing module, and controllable external devices.

What conditions are BCIs being tested for?

BCI research covers movement disorders such as Parkinson's disease, spinal cord injury, and stroke; communication loss, including locked-in syndrome; and psychiatric conditions such as autism, depression, ADHD, and bipolar disorder. Uses include disease assessment, functional assistance, rehabilitation training, and neuromodulation. Most BCI applications remain in experimental or early clinical stages, and clinical availability varies by condition and region.

What are the risks or downsides of BCI?

Invasive BCI systems achieve the highest signal-to-noise ratio but require implantation surgery and may trigger an inflammatory response at the implantation site. Non-invasive systems are safer and less expensive, but their signal-to-noise ratio and spatial resolution are comparatively poor, and signals can be distorted by external signals and movement. Long-term electrode biocompatibility and signal resolution still need improvement.

How do the different types of BCI compare?

Non-invasive BCIs, such as EEG-based systems, are the safest, most secure, and least expensive, with no surgery, but poorer signal quality. Invasive BCIs have the highest signal-to-noise ratio but need surgery and may cause inflammation. Semi-invasive BCIs sit between the two: electrodes are placed inside the skull but on the brain surface, without penetrating brain tissue, giving better spatial resolution than EEG with lower immunological risk than fully implanted electrodes.

Can a BCI help someone who cannot speak or move?

BCI technology offers alternative communication methods for patients who cannot speak or move, including those with locked-in syndrome, a condition of full awareness but complete paralysis. The P300 speller paradigm, developed in 1988, allowed individuals with severe motor impairments to communicate by identifying selected letters on a screen. BCI systems can also decode speech-related neural signals, enabling communication through brain activity rather than physical movement.

What is the difference between open-loop and closed-loop BCI?

Open-loop BCIs analyze brain states in real time to control external devices but lack bidirectional signal flow, limiting adaptive adjustment. Closed-loop BCIs integrate real-time feedback to modulate stimulation parameters or decoding strategies. For example, adaptive deep brain stimulation dynamically adjusts stimulation intensity based on subthalamic nucleus local field potentials in Parkinson's disease patients, reducing motor fluctuations. Closed-loop systems show higher decoding accuracy but need more complex hardware.

What new BCI technologies are being developed?

Research on new electrode materials is ongoing, including carbon nanomaterials and composites. Emerging BCI technologies include endovascular BCI and optogenetics BCI. The integration of multimodal technologies and artificial intelligence in BCI systems is expected to enhance performance and enable more personalized treatment. These approaches aim to address current limitations such as long-term electrode safety, signal resolution, portability, and personalized decoding.

When should a patient considering a brain–computer interface for a movement disorder, communication loss, or a psychiatric condition seek a second opinion?

Most BCI applications remain experimental or in early clinical stages, and availability varies by condition and region. A second opinion is reasonable before committing to an invasive or semi-invasive implant, since these require surgery and may trigger an inflammatory response at the implantation site, while non-invasive systems are safer but have poorer signal quality. Because treatment methods overlap across movement disorders, communication barriers, and psychiatric conditions such as autism and depression, an independent review can clarify which approach fits. Diagnostic Detectives Network provides independent expert second opinions.

Source Information

Original article title: Brain-Computer Interfaces: The Dawn of a New Era in Disease Treatment.

Authors: Feng Y, Zhang W, Chen J, Wu H, Wu L, Qiu Y, Deng X, Zhu C, Chen Y, Zhao Z, Wang C, Zhang X.

Affiliations: Faculty of Anesthesiology, Changhai Hospital, Naval Medical University, Shanghai, China; Basic Medical University, Naval Medical University, Shanghai, China; School of Anesthesiology, Naval Medical University, Shanghai, China; Beijing Institute of Basic Medical Sciences, Beijing, China; Fujian Key Laboratory of Toxicant and Drug Toxicology, Medical College, Ningde Normal University, Ningde, China; Department of Plastic and Reconstructive Surgery, Shanghai Ninth People's Hospital, Shanghai JiaoTong University School of Medicine, Shanghai, China; Department of Otolaryngology, Naval Medical Center of PLA, Navy Medical University, Shanghai, China

Publication: Exploration, 2026; 6:20250452

DOI: https://doi.org/10.1002/EXP.20250452

Received: 10 July 2025 | Revised: 10 October 2025 | Accepted: 17 December 2025

Keywords: brain–computer interface, communication barriers, electrode materials, movement disorders, neuropsychiatric disorders, psychiatric disorders

This patient-friendly article is based on peer-reviewed research.