Application of brain-computer interfaces in the rehabilitation of post-stroke emotional and sleep disorders

Abstract

Post-stroke emotional and sleep disorders are non-motor sequelae that can interfere with rehabilitation, functional recovery, and quality of life. Brain-computer interfaces (BCIs) have gained increasing attention in post-stroke rehabilitation research, especially for assessing and facilitating motor functions based on the translation and decoding of electroencephalogram signatures with closed-loop feedback. Although direct evidence supporting BCI-based interventions for post-stroke emotional and sleep disorders is limited, emerging evidence indicates that BCIs may extend beyond motor rehabilitation by improving active participation, neuroplasticity, and self-regulation. Although current evidence remains indirect, it is derived specifically from studies on motor rehabilitation, as well as from multimodal interventions and related fields such as neurofeedback and mindfulness-based rehabilitation. In particular, BCI-based strategies combined with psychological interventions, electrical stimulation, or immersive feedback approaches may open new perspectives for integrated intervention in both motor and non-motor domains, which significantly impact quality of life. Accordingly, this minireview summarizes the current evidence, theoretical rationale, and future directions for BCI applications in the rehabilitation of post-stroke emotional and sleep disorders. BCIs should be viewed as an emerging and potentially valuable platform for multidimensional post-stroke rehabilitation rather than as an established standalone treatment; thus, their clinical role warrants further investigation.

Key Words: Brain-computer interface; Stroke rehabilitation; Post-stroke depression; Post-stroke anxiety; Sleep disorders; Neurofeedback; Closed-loop rehabilitation

Core Tip: This minireview highlights that brain-computer interfaces (BCIs), primarily studied for post-stroke motor rehabilitation, may also address post-stroke emotional and sleep disorders. Although direct evidence is limited, indirect findings suggest that BCIs may enhance active participation, neuroplasticity, and self-regulation. Combining BCIs with psychological, electrical, or immersive feedback approaches offers new integrative perspectives. BCIs represent an emerging multidimensional rehabilitation platform, not an established standalone treatment, warranting further clinical research.

INTRODUCTION

Despite advances in healthcare, stroke remains the leading cause of long-term disability, imposing a substantial burden on both patients and the society. Therefore, post-stroke rehabilitation has become a serious health concern worldwide, and is mainly focused on motor recovery, such as the restoration of upper limb function and activities of daily living. However, because many stroke survivors experience persistent non-motor complications, including depression, anxiety, and sleep disorders, which may reduce motivation, limit rehabilitation participation, and negatively influence functional outcomes[1], the need for comprehensive rehabilitation strategies has become increasingly urgent.

Post-stroke emotional disorders are particularly relevant because they are frequently and clinically impactful. Approximately 20% of patients are reported to be anxious after experiencing a stroke, with pooled anxiety rates of 20% within 1 month, 23% at 1-5 months, and 24% at 6 months after stroke[2]. An updated systematic review also reported anxiety in approximately one-quarter of patients when determined by rating scales, and one-fifth of patients when determined through clinical interviews[3]. Post-stroke depression has also been reported as a strong predictor of worse functional outcomes, further highlighting the importance of incorporating psychological evaluations in post-stroke rehabilitation[4].

Sleep disorders also contribute substantially to non-motor impairments in stroke survivors, including symptoms of insomnia, poor sleep quality, and circadian rhythm disturbances, which may be present after stroke and influence neurological recovery, emotional well-being, and quality of life[5]. Consistent with previous findings, recent reviews have indicated that sleep disorders after stroke are associated with functional prognosis, rehabilitation response, and quality of life[6]. Collectively, these results highlight that sleep and emotional outcomes should not be viewed as secondary outcomes but rather as key constructs of post-stroke recovery.

Brain-computer interfaces (BCIs) have been proposed as an advantageous approach to neurorehabilitation, even in the early stages. A BCI system measures the brain activity, extracts pertinent features, decodes motor intentions or brain states, and delivers real-time feedback through visual, auditory, robotic, electrical, or virtual modalities. Most BCI studies investigating stroke rehabilitation have focused on motor recovery, particularly upper-limb rehabilitation, using electroencephalogram (EEG)-based motor imagery paradigms. A review by Cervera et al[7] summarized the evidence supporting the clinical efficacy of post-stroke motor rehabilitation using BCI and provided a basis for this field. More recently, there has been evidence that BCI-based rehabilitation may contribute to motor recovery, although there is an overall large variation in the study design, patient selection, feedback modality, and intervention dose across studies[8].

The applicability of BCIs to post-stroke emotional and sleep disturbances is not yet clear; however, several of their characteristics hold promise for non-motor rehabilitation. First, BCIs enable a closed-loop feedback that connects brain activity to external sensory or behavioral feedback, potentially facilitating neuroplasticity and active learning. Second, BCI-based training involves the active participation of patients, which may enhance cohesion, self-efficacy, and compliance with rehabilitation. Third, EEG neurofeedback and similar closed-loop processes can facilitate training of arousal, attention, emotions, and sleep-related brain states. Fourth, BCIs can be integrated with mindfulness techniques, functional electrical stimulation (FES), robotics, virtual reality, and home care to develop comprehensive rehabilitation models. Notably, a randomized controlled trial of BCIs combined with mindfulness therapy in patients with hemiplegic stroke demonstrated enhanced motor function, activities of daily living, mindful attention awareness, sleep quality, and quality of life, offering the first glimpse of the association of BCIs with more extensive rehabilitation outcomes[9].

Accordingly, this minireview highlights the current evidence, theoretical underpinnings, potential mechanisms, challenges, and future directions of BCI technology applications in the rehabilitation and management of post-stroke emotional and sleep disorders. Based on the available evidence, this minireview does not imply that BCIs are effective interventions for post-stroke depression, anxiety, or sleep disorders. Instead, it reviews the potential of BCIs in motor rehabilitation to be extended to holistic neuropsychological rehabilitation with a focus on emotional regulation, sleep quality, patient engagement, and closed-loop personalized care.

LITERATURE SEARCH STRATEGY

We performed a narrative literature search of the PubMed database for relevant studies published between its inception and March 2026. Search terms included the following: “brain-computer interface”, “brain computer interface”, “BCI”, “EEG-based BCI”, “motor imagery”, “neurofeedback”, “closed-loop rehabilitation”, “functional electrical stimulation”, “robotic rehabilitation”, “virtual reality”, “stroke”, “post-stroke rehabilitation”, “post-stroke depression”, “post-stroke anxiety”, “sleep disorder”, “insomnia”, “sleep quality”, and “sleep-disordered breathing”.

Priority was given to systematic reviews, meta-analyses, randomized controlled trials, and clinically relevant mechanistic studies. Because direct studies investigating BCI for post-stroke emotional and sleep disorders remain limited, adjacent evidence from neurofeedback, mindfulness-based rehabilitation, virtual reality, and sleep-related feedback interventions have also been considered when clarifying potential mechanisms or future applications. This minireview used a narrative design and did not aim to provide a quantitative synthesis.

POST-STROKE EMOTIONAL AND SLEEP DISORDERS

Post-stroke depression and anxiety

Post-stroke depression and anxiety are among the most frequent emotional complications of stroke. In this minireview, post-stroke emotional disorders mainly refer to depression and anxiety, which are the most commonly studied emotional complications of stroke rehabilitation research. They may arise in the acute, subacute, or chronic stages, and persist well beyond the point of neurological stabilization. These disorders are influenced by multiple factors, including lesion characteristics, neurological severity, functional disability, cognitive impairment, social support, premorbid psychiatric history, inflammatory processes, and the subjective experience of loss of independence[10-12].

Although post-stroke anxiety is underappreciated, it is common. Excessive worry about recurrence, fear of falling, avoidance of rehabilitation exercises, irritability, muscle tension, sleep disturbances, and diminished confidence in daily life are some manifestations. Campbell Burton et al[2] reported that anxiety remains a common problem at different time points after stroke, with pooled estimates of 20%-24%. Knapp et al[3] found that a significant proportion of stroke survivors experienced anxiety; however, the prevalence depended on whether a rating scale or diagnostic interview was used. These results are clinically relevant because anxiety may hinder participation in rehabilitation and decrease the propensity to perform demanding motor and cognitive tasks.

Depression after stroke may also have powerful effects on recovery[4], increase fatigue, impaired sleep, and reduced adherence to rehabilitation[13]. It is also associated with poor functional outcomes, reinforcing the need for its early identification and integrated management during rehabilitation[14,15]. Nevertheless, it may be difficult to consistently perform these interventions in some stroke survivors, particularly those with communication, cognitive, or severe motor impairments. Limited access to trained therapists and reduced clinical contact after structured rehabilitation may further hinder long-term implementation.

From the standpoint of BCI-based rehabilitation, post-stroke depression and anxiety are pertinent from the perspective of BCI-based rehabilitation for 2 reasons. First, emotional symptoms may hinder motor rehabilitation by reducing attention and motivation to participate in rehabilitation as well as the intensity of training. Second, the BCI may impact emotional outcomes through indirect pathways, such as increasing motor control, active participation, and self-efficacy by restoring agency. Thus, it may be more appropriate to consistently incorporate emotional outcomes into future BCI trials, even though motor recovery continues to be the primary endpoint.

Post-stroke sleep disorders

Post-stroke sleep disorders are heterogeneous and include insomnia, poor sleep quality, and other symptoms. Sleep-disordered breathing should be distinguished from insomnia and subjectively poor sleep quality, because it involves different pathophysiological mechanisms, assessment methods, and treatment strategies[5,6]. These may be the consequences of direct brain damage, sleep-wake network disruption, immobility, pain, medication, mood disorders, hospitalization, environment, and coexisting cardiometabolic disease(s).

The clinical relevance of post-stroke sleep disorders goes well beyond nighttime complaints, which may exacerbate fatigue during the day, as well as cognitive and emotional functioning. Sleep-disordered breathing also contributes to an increased vascular risk and may lead to recurrent cerebrovascular events[1]. A review by Cai et al[5] pointed out that sleep disorders are prevalent among individuals with cerebrovascular diseases, including insomnia, hypersomnia, and breathing-related diseases, and are often characterized by common mental and behavioral disturbances that potentially limit functional prognosis, rehabilitation outcomes, and quality of life[6]. The following subjective questionnaires are most commonly used in clinical studies: The Pittsburgh Sleep Quality Index, the Insomnia Severity Index, and the Epworth Sleepiness Scale. Sleep diaries, polysomnography, actigraphy, and wearable devices can provide more objective data. However, a comprehensive sleep evaluation in rehabilitation research is lacking. This is a limitation of BCI research because sleep quality may affect attention, cortical excitability, motor learning, and the ability to perform motor imagery or neurofeedback tasks.

BCI-based sleep rehabilitation after stroke remains largely exploratory, and direct evidence that BCI improves post-stroke insomnia and sleep quality is limited. Nevertheless, EEG-based BCI and neurofeedback are technically relevant because sleep and arousal states can be reflected in brain rhythms. However, evidence for sleep-related neurofeedback remains unclear. A systematic review and meta-analysis[16] found that surface neurofeedback does not provide additional benefits over controlled conditions in terms of self-perceived sleep quality or insomnia. Future studies should investigate whether EEG-guided feedback, relaxation-oriented BCI, closed-loop monitoring, and wearable sleep technologies can be integrated into rehabilitation programs for stroke patients.

Interaction between emotional and sleep disturbances

Post-stroke emotional disturbances and sleep disturbances are mutually associated, and sleep deprivation may worsen the symptoms of depression and anxiety by contributing to fatigue, impeding emotional regulation, and diminishing cognitive control. Depression and anxiety can increase the severity of insomnia, deepen night-time rumination, disrupt sleep continuity, and decrease perceived sleep quality. This bidirectional relationship may create a vicious cycle in which emotional and sleep disturbances reinforce each other, thereby reducing rehabilitation engagement and slowing functional recovery[17,18]. This interplay is particularly salient in rehabilitation using BCI. For example, motor-imagery BCI, neurofeedback, and closed-loop systems require sustained attention and learning, and repeated training is required. However, as BCI-based rehabilitation enhances motor function, self-efficacy, relaxation, and the regulation of brain states, it may have an indirect effect on emotional well-being and sleep. Therefore, emotional and sleep-related outcomes in future BCI trials should not be viewed as mere secondary observations but rather as factors that may interact with the rehabilitation response. Therefore, future BCI research should assess baseline emotional and sleep statuses and examine whether these characteristics moderate the feasibility, learning, and clinical response to BCI training.

PRINCIPLES OF AND REHABILITATION RATIONALE FOR BCIs

Basic components of BCI systems

A BCI enables communication or control through brain signals rather than using conventional neuromuscular output. In rehabilitation, BCI is commonly used not only for external control but also for therapeutic feedback. Typical BCI systems include signal acquisition, preprocessing, feature extraction, classification or decoding, and feedback delivery[19]. Brain signals can be recorded using invasive or noninvasive methods. However, noninvasive EEG-based BCI is the most widely used approach for stroke rehabilitation because of its relatively low cost, portability, safety, and high temporal resolution[20].

In EEG-based motor imagery BCI, patients are instructed to imagine the movement of the affected limb[21,22]. The system monitors event-related desynchronization and synchronization patterns in the sensorimotor rhythms and converts these patterns into feedback. Feedback can be delivered on a computer screen through virtual limbs, robots, FES, or other sensory inputs. Online feedback enables patients to link attempted or imagined movements to external sensory outcomes, thereby closing the sensorimotor loop.

BCI-based rehabilitation differs from traditional repetitive training in that it focuses on the connections between intention, brain activation, and feedback. This is particularly important for patients with significant motor deficits, because they may have difficulty generating adequate voluntary motion. By using BCI-based rehabilitation to directly decode motor intentions from brain signals, these patients can be actively involved in rehabilitation although with limited overt movements.

EEG-based BCI, motor imagery, neurofeedback, and closed-loop systems

In this minireview, BCI refer to systems in which brain-derived signals are recorded, processed, and used to provide feedback or control an external device. EEG neurofeedback is considered a related but not identical approach, because it emphasizes the self-regulation of brain activity rather than external device control. Virtual reality and mindfulness are not BCI technologies by themselves; they are discussed in this minireview only when they are combined with BCI or when they provide adjacent evidence for non-motor rehabilitation. Similarly, closed-loop neuromodulation shares the principle of adaptive feedback, but should not be considered synonymous with BCI in all contexts. Several BCI paradigms have been investigated for use in post-stroke rehabilitation. The motor-imagery-based BCI is the most established paradigm. Patients imagine the movements of their paretic limb while EEG signals are decoded and converted into feedback to reinforce sensorimotor activation and promote neuroplasticity[7,23-25].

BCI-controlled FES links motor intention to electrical stimulation of the affected muscles[26]. This approach may strengthen the association between the cortical activation and peripheral sensory feedback. A recent systematic review evaluated BCI-FES training for upper limb functional recovery after stroke, highlighting its therapeutic potential, while also noting substantial variations in stimulation parameters, mental tasks, and threshold settings[27,28].

BCI-assisted robotic rehabilitation combines brain signal decoding with robotic movement assistance[29], in which robotic devices provide high-intensity, repetitive, and task-specific training, whereas BCI may ensure that assistance is contingent on a patient’s motor intention[30]. This contingency may be important, because feedback linked to the patient’s brain activity may better support learning and agency than passive movements alone.

Virtual environments can simulate functional tasks, provide real-time visual reinforcement, and increase patient engagement[31]; as such, BCI incorporating virtual reality may provide immersive, motivating, and multisensory feedback[32]. This direction may be particularly relevant for emotional rehabilitation because immersive feedback and task success may improve motivation, confidence, and enjoyment during training[33]. However, evidence from virtual reality-based rehabilitation should be interpreted as adjacent, rather than direct, evidence for BCIs unless the virtual environment is driven by brain-derived signals.

Both EEG neurofeedback and closed-loop BCI are concerned with the control of brain states and not simply with movement intentions. Neurofeedback provides users with immediate data on brain activity, and teaches them how to change specific neural patterns. An extensive review of neurofeedback has detailed its application in several clinical conditions, including insomnia, anxiety, and depression[34]. Although such evidence is not stroke specific, it provides a theoretical bridge between BCI technology and emotional or sleep-related rehabilitation.

Rationale for extending BCI from motor to non-motor rehabilitation

There are several reasons for the expansion of the application of BCI, from motor recovery to emotional and sleep rehabilitation. Stroke recovery is a complex and multifaceted process. Motor impairment, emotional distress, sleep disruptions, fatigue, and cognitive impairment may also occur. Rehabilitation technology that enhances motor outcomes but disregards psychological and sleep-related issues will likely overlook the comprehensive recovery needs of stroke survivors.

Second, BCIs are more than motor-assistive technologies; they are brain state-driven feedback systems. Thus, BCIs can be designed, in principle, to identify and modulate motor imagery signals as well as the neural correlates of attention, arousal, relaxation, fatigue, and emotional regulation. However, the corresponding strategies for these specific applications are still in their infancy. EEG neurofeedback has been studied as a potential treatment for depression, anxiety, and insomnia in non-stroke populations[35,36]. This finding supports the conceptual possibility of adapting BCI-related methods for post-stroke non-motor rehabilitation.

Third, BCIs may have secondary effects on emotional and sleep parameters through positive changes in agency and self-efficacy. Many stroke survivors experience frustration, helplessness, and reduced confidence owing to motor disabilities. Through BCI training, invisible intentions can become visible with the help of external feedback, enabling patients to feel a sense of control even when motion is restricted. This restoration of agency may be psychologically meaningful and enhance rehabilitation motivation.

Fourth, BCI may be applied in combination with interventions that directly affect emotional and sleep-related outcomes. Mindfulness training, relaxation methodologies, cognitive-behavioral approaches, virtual reality, and remote rehabilitation platforms can be combined with BCI feedback. A randomized controlled trial in which BCI was combined with mindfulness-based therapy in hemiplegic patients with stroke was also noteworthy because it connected BCI-based rehabilitation with sleep quality and quality of life results[9].

The rationale for applying BCI to post-stroke emotional and sleep rehabilitation should be regarded as an emerging translational concept rather than an established treatment claim. The present findings support BCI as a promising platform for motor rehabilitation and suggest several plausible pathways for non-motor benefits. Nevertheless, direct clinical evidence addressing post-stroke depression, anxiety, and sleep disorders remains limited, and future investigations should focus on BCI-based emotional and sleep rehabilitation using standardized outcomes, appropriate control groups, and long-term follow-up.

Current evidence supporting BCI-based interventions

This minireview divides the representative direct, indirect, and adjacent evidence into three levels, as summarized in Table 1. The first provides direct evidence such as studies reporting BCI-based stroke rehabilitation with emotional, sleep-related, or quality-of-life outcomes. The second level includes indirect evidence such as BCI-based motor rehabilitation trials and meta-analyses, which may affect emotional or sleep-related recovery by enhancing motor function, agency, and participation. The third is the adjacent level, including neurofeedback, virtual reality, mindfulness-based rehabilitation, and sleep-related feedback interventions that are not always BCI-based or stroke-specific but offer mechanistic or translational evidence.

Table 1 Evidence map of direct, indirect, and adjacent evidence for brain-computer interfaces-related post-stroke emotional and sleep rehabilitation.

Evidence type	Population	Intervention	Outcomes	Relevance to this review	Ref.
Direct evidence	Hemiplegic patients after stroke	BCI combined with mindfulness therapy	Motor function, ADL, mindful attention, sleep quality, quality of life	Early evidence linking BCI-based rehabilitation with sleep and quality-of-life outcomes	Wang et al[9], 2023
Indirect BCI evidence	Stroke survivors	BCI-based motor rehabilitation	Motor recovery	Supports BCI as a motor rehabilitation platform; emotional and sleep effects remain indirect	Cervera et al[7], 2018
					Tonin et al[8], 2025
					Mortezaei et al[37], 2026
Representative BCI trials	Stroke survivors	MI-BCI, BCI-robot, BCI-exoskeleton	Upper-limb motor function	Demonstrates feasibility of intention-driven feedback-based rehabilitation	Ang et al[23], 2015
					Pichiorri et al[24], 2015
					Frolov et al[25], 2017
BCI-FES evidence	Stroke survivors	BCI-controlled FES	Upper-limb recovery	Supports closed sensorimotor feedback and neuroplasticity rationale	Ren et al[26], 2024
					Zhang et al[38], 2025
Adjacent VR evidence	Stroke survivors	VR-based rehabilitation	Motor function, depression, participation	Supports immersive and motivational feedback but is not direct BCI evidence	Laver et al[31], 2025
					Wei et al[32], 2025
					Liu et al[33], 2023
Adjacent neurofeedback evidence	Non-stroke or mixed populations	Electroencephalogram neurofeedback	Depression, anxiety, insomnia, sleep quality	Provides theoretical bridge for brain-state self-regulation	Marzbani et al[34], 2016
					Patil et al[36], 2023
					Recio-Rodriguez et al[16], 2024

Direct evidence refers to brain-computer interface-based stroke rehabilitation studies reporting emotional, sleep-related, or quality-of-life outcomes. Adjacent evidence refers to related neurofeedback, virtual reality, mindfulness-based rehabilitation, or sleep-feedback studies that provide mechanistic or translational support but do not directly establish brain-computer interface efficacy for post-stroke emotional or sleep disorders.

BCI: Brain-computer interfaces; ADL: Activities of daily living; MI: Motor imagery; FES: Functional electrical stimulation; VR: Virtual reality.

Direct evidence for emotional and sleep-related outcomes

Direct evidence for BCI-based interventions targeting post-stroke emotional and sleep disorders remains limited. Most available BCI trials for stroke rehabilitation have been designed primarily to improve motor function; only a small proportion have assessed sleep quality, emotional symptoms, and quality of life as secondary outcomes. Among the available studies, the randomized controlled trial by Wang et al[9] was particularly relevant because it evaluated BCI combined with mindfulness therapy in patients with hemiplegia after stroke and reported improvements in motor function, activities of daily living, mindful attention awareness, sleep quality, and quality of life. These results suggest that BCI intervention techniques can be combined with psychological self-management strategies to achieve a broader recovery. However, as the intervention combined BCI with mindfulness therapy and rehabilitation care, the independent contributions of BCI to sleep and emotional improvement could not be isolated.

Existing studies support the feasibility of incorporating emotional and sleep-related outcomes into BCI-based stroke rehabilitation; however, BCI has not yet been validated as a standalone intervention for post-stroke depression, anxiety, or sleep disorders. Therefore, it is important to distinguish clinical observations from mechanistic hypotheses. At present, clinical evidence indicates that BCI-based or BCI-combined rehabilitation can improve motor outcomes and, as per a limited number of studies, may be associated with improvements in sleep quality or quality of life. In contrast, the proposed effects of agency, self-efficacy, motivation, and active participation on emotional symptoms remain largely as mechanistic hypotheses rather than confirmed longitudinal pathways. Future studies should examine whether changes in self-efficacy, perceived control, engagement in rehabilitation, and sleep quality mediate the relationship between BCI-based training and emotional outcomes.

Indirect evidence from BCI-based motor rehabilitation

Most published studies on BCI-based stroke rehabilitation have investigated motor recovery rather than emotional or sleep outcomes. In these studies, BCI was generally used to decode motor imagery or attempted movement from EEG signals and then deliver contingent feedback through visual displays, robotic assistance, FES, or virtual environments[37]. By linking the motor intention to external feedback, BCI-based training aims to reinforce sensorimotor activation and promote activity-dependent neuroplasticity[37].

A meta-analysis by Cervera et al[7] provided early quantitative evidence supporting the clinical effectiveness of BCI-based post-stroke motor rehabilitation. The authors evaluated controlled trials and concluded that BCI interventions demonstrated beneficial effects on motor recovery compared to control interventions, while emphasizing the need for better-designed clinical studies. Later reviews continued to support the potential of non-invasive BCI-assisted rehabilitation as an adjunct to conventional therapy, but they also highlighted important sources of heterogeneity, including stroke stage, lesion characteristics, motor impairment severity, BCI paradigm, feedback modality, training dose, and outcome measures[8,38]. Several representative trials have contributed to this finding. Ang et al[23] investigated EEG-based motor imagery BCI combined with robotic rehabilitation in stroke patients, whereas Pichiorri et al[24] reported that BCI-supported motor imagery training could enhance motor imagery practice during stroke recovery. Frolov et al[25] further evaluated motor-imagery-based BCI-controlled hand exoskeleton training in a randomized multicenter trial. Although these studies do not directly support BCI as a therapy for post-stroke depression, anxiety, or sleep disturbances, they remain relevant because motor recovery is related to mood, independence, and self-efficacy[39]. In this sense, BCI-based motor rehabilitation may have psychosocial implications even when emotional and sleep-related outcomes are not primary targets. However, these potential indirect benefits remain insufficiently quantified largely because many BCI trials have not included standardized measures of depression, anxiety, and other aspects.

BCI-FES and BCI-robotic rehabilitation as closed sensorimotor approaches

The BCI combined with FES is one of the most clinically relevant paradigms for stroke rehabilitation. In this model, the patient’s motor intention or motor imagery-related EEG activity triggers electrical stimulation of the paretic muscles. This creates a closed loop between the cortical activation, peripheral movement, proprioceptive feedback, and sensory reafference. Theoretically, this synchronization may strengthen sensorimotor pathways and enhance motor relearning.

A recent systematic review assessed the impact of BCI-based FES on upper limb functional recovery after stroke[26]. The review concluded that BCI-FES training may have a therapeutic effect; however, it also pointed out that the stimulation protocols, training tasks, mental strategies, BCI classifiers, and threshold levels varied greatly across studies. A network meta-analysis comparing BCI-FES, transcranial direct current stimulation, and conventional motor rehabilitation also suggested that BCI-FES may be a promising approach for upper limb recovery; however, further high-quality evidence is needed[40].

BCI-assisted robotic rehabilitation follows the closed-loop principle. Robotic devices provide repetitive, task-specific, and measurable movement assistance, whereas BCI ensure that assistance is contingent on a patient’s brain activity rather than being passively delivered. This contingency is clinically meaningful because it may enhance the patient’s sense of agency and active involvement. From the perspective of emotional recovery, having agency matters; patients who believe that their intention can result in a visible movement or feedback may feel more empowered and less helpless. However, BCI-robotic studies have not systematically assessed psychological or sleep outcomes; consequently, an important knowledge gap remains.

Adjacent evidence from mindfulness, virtual reality, and neurofeedback

Mindfulness, virtual reality, and neurofeedback offer parallel rather than conclusive evidence supporting the effectiveness of BCI in post-stroke emotional and sleep-related disorders. These methods are pertinent because they have similar foci of attention, self-regulation, motivation, relaxation, or immersive feedback and may be adapted for use with BCI systems in future rehabilitation paradigms. Mindfulness practice integrated with a BCI is particularly applicable to post-stroke emotional and sleep recovery. Mindfulness-based treatments are designed to promote improved regulation of attention, emotional awareness, stress appraisal, and relaxation. Mindfulness practices paired with a BCI can incorporate brain state feedback with psychological self-regulation. In a randomized controlled trial of BCI plus mindfulness therapy in patients with hemiplegia after stroke, Wang et al[9] reported improvements in motor function, activities of daily living, mindful attention awareness, sleep quality, and quality of life.

Virtual reality is another promising adjunct to BCI-based rehabilitation that provides immersive, task-oriented, and motivating feedback. Therefore, BCI-based rehabilitation using virtual reality may increase the attention, engagement, and enjoyment during training. In addition, virtual reality has been used to support motor training, balance training, and functional task simulation. When integrated with a BCI, virtual reality can provide real-time visual feedback based on motor imagery or brain-state changes. This integration may be useful for patients with emotional symptoms, because meaningful task success in a virtual environment may improve motivation and self-efficacy. However, the current evidence for virtual reality-based improvement in post-stroke depression should be interpreted as supportive adjacent evidence rather than direct evidence for BCI unless the virtual reality feedback is explicitly driven by brain signals. The effects of neurofeedback on anxiety, depression, insomnia, and related symptoms have been investigated[34-36,41]. However, in stroke, EEG neurofeedback may be relevant not only for motor recovery, but also for attention, emotional control, and sleep-related brain states.

Potential applications in post-stroke sleep rehabilitation

BCI-based sleep rehabilitation after stroke remains an emerging concept rather than an established clinical approach. Sleep and arousal states are reflected in the EEG rhythms, which provide a theoretical basis for EEG-guided monitoring, neurofeedback, and closed-loop modulation. However, most studies investigating neurofeedback during sleep are not stroke-specific. Therefore, the application of BCI for sleep post-stroke should be positioned as a future translational direction rather than a currently established clinical tool.

Different post-stroke sleep phenotypes may require different rehabilitation strategies. EEG neurofeedback or relaxation-oriented BCI may be more relevant to insomnia, hyperarousal, and subjective poor sleep quality, because these conditions are closely related to arousal regulation, sleep perception, and cortical rhythm modulation. In contrast, sleep-disordered breathing has a different pathophysiological basis and usually requires airway or respiratory support rather than BCI-based interventions alone. Circadian rhythm disturbances may be more closely related to behavioral scheduling, light exposure, daytime activity, and sleep-wake irregularity. Therefore, future BCI-based sleep rehabilitation strategies should not treat post-stroke sleep disorders as a single, uniform entity. Instead, BCI should be matched to specific sleep phenotypes and positioned as an adjunctive strategy when brain-state feedback or self-regulation is mechanistically relevant.

A systematic review and meta-analysis of the effects of neurofeedback on sleep quality and insomnia revealed that surface neurofeedback did not convincingly outperform control conditions for self-rated sleep quality and insomnia symptoms[16]. This is a crucial finding of the current review, because it prevents exaggeration. Neurofeedback and BCI-based methodologies appear to be promising; however, their impact on sleep is inconclusive and is considerably influenced by the study type, training protocols, control conditions, and outcome measures. Future BCI-based applications for post-stroke sleep may encompass EEG-based relaxation training; brain-state feedback for arousal regulation; closed-loop sleep monitoring; integration with wearable technology; and multimodal intervention with cognitive behavioral therapy for insomnia, breathing training, mindfulness, or physical therapy. These findings may be particularly applicable to home or remote treatment of stroke survivors. However, protocols for future clinical trials should characterize the target sleep phenotype and use validated instruments, such as the Pittsburgh Sleep Quality Index, Insomnia Severity Index, Epworth Sleepiness Scale, actigraphy, and polysomnography, if applicable.

POTENTIAL MECHANISMS

Neuroplasticity and brain network reorganization

A conceptual model connecting BCI-based rehabilitation strategies to potential outcomes in the motor, emotional, sleep-related, and quality of life domains is shown in Figure 1. A stroke interferes with the neural circuitry that governs motor control, cognition, emotion, and arousal. BCI training can facilitate reorganization through repetitive associations between motor intentions or brain state modulations with external feedback. Event-related desynchronization or synchronization in sensorimotor rhythms is also used to recognize imagined movement in motor imagery BCI[24]. Once the detected brain activity leads to visual feedback, robotic movement, or FES, the patient receives sensory feedback that is temporally connected to cortical activation. This closed sensorimotor loop facilitates functional neural rewiring[42].

Figure 1 Conceptual framework of brain-computer interface-related rehabilitation for post-stroke emotional and sleep disorders. Brain-computer interface-related approaches may support integrated post-stroke rehabilitation by linking emotional symptoms, sleep disturbance, fatigue, and participation with potential motor and non-motor outcomes through closed-loop feedback, neuroplasticity, and self-regulation. The pathways shown are proposed mechanisms because direct evidence for emotional and sleep-related outcomes remains limited. Virtual reality and neurofeedback provide adjacent evidence unless directly driven by brain signals. BCI: Brain-computer interfaces; MI: Motor imagery; FES: Functional electrical stimulation; EEG: Electroencephalogram; VR: Virtual reality.

While this mechanism has mainly been studied in motor rehabilitation, it may also be applicable to the recovery of emotional and sleep processes, such as the regulation of the prefrontal, limbic, thalamocortical, and brainstem regions. Although BCI-related feedback may not directly repair these networks, it can modulate cortical excitability and patient participation, which, in turn, could facilitate a more generalized recovery. However, it remains controversial whether these neuroplastic adaptations also ameliorate post-stroke depression, anxiety, or sleep disturbances.

Neuropsychiatric mechanisms relevant to emotion and sleep

Beyond sensorimotor neuroplasticity, emotional and sleep-related rehabilitation requires the consideration of neuropsychiatric mechanisms. Post-stroke emotional symptoms are closely related to distributed networks involving prefrontal regulation, limbic reactivity, cognitive control, and reward processing. Sleep and arousal regulation also depend on the thalamocortical and brainstem-related networks. From this perspective, BCI-related approaches may be relevant not only because they provide motor feedback but also because EEG-based feedback and neurofeedback can theoretically target brain-state regulation. Potential neurophysiological markers include frontal asymmetry, α-band activity, sensorimotor rhythm modulation, and connectivity-related features. However, these markers have not been sufficiently validated as therapeutic targets for post-stroke depression, anxiety, or sleep disorders. Therefore, the neuropsychiatric mechanisms proposed in this minireview should be interpreted as plausible translational pathways rather than established mechanisms of clinical efficacy.

Sensorimotor recovery and emotional regulation

Motor impairment after stroke can lead to frustration, reliance on others for activities of daily living, social isolation, and loss of self-esteem. This can lead to depression, anxiety, or poor sleep quality. BCI-based rehabilitation may help patients regain a sense of control by providing evidence that brain activity or motor intentions can lead to the generation of external feedback. This finding is particularly significant in patients with profound paresis who are unable to perform overt movements[4,43]. Therefore, the psychological impact of an agency should not be overlooked. In traditional passive interventions, patients have the impression that they are moving. In BCI-based training, successful feedback is determined by the patient’s brain activity, which may further strengthen the feelings of active involvement. This may enhance self-efficacy and the willingness to participate in rehabilitation. These psychosocial effects are likely, in part, responsible for the potential effects of BCI-based interventions on the quality of life, motivation, and sleep. However, the proposed mechanism is hypothetical, and should be directly assessed in future studies.

EEG neurofeedback and self-regulation

EEG neurofeedback regulates specific neural patterns and has been studied in non-stroke populations for anxiety, depression, and other stress-related symptoms[35]. The association between neurofeedback and post-stroke emotional and sleep disorders lies in the possibility that patients learn to regulate their arousal, attention, and emotional reactivity through repeated feedback[34]. For example, neurofeedback schemes may aim at alpha- (α), theta- (θ), or beta- (β) band activity or sensorimotor rhythms depending on the desired clinical outcome. In sleep-related applications, neurofeedback has traditionally focused on sensorimotor rhythm increase or relaxation-related brain activity, while emotional regulation programs may influence frontal asymmetry or α-band activity or connectivity-related features. However, these protocols have not yet been standardized for stroke survivors because stroke-related neurological deficits, fatigue, and altered EEG patterns may affect neurofeedback learning[41].

Sleep-emotion-rehabilitation interaction

As previously described, sleep, emotion, and rehabilitation outcomes are interrelated, and emotional distress may affect sleep through mechanisms such as rumination, hyperarousal, and decreased in daytime activity. This reciprocal association can hinder participation in rehabilitation and restrict functional recovery[17,18]. BCI-based rehabilitation may affect this cycle via several plausible pathways. First, BCI training enhances motor function and independence, thereby reducing emotional distress and supporting sleep recovery[7,8]. Furthermore, a BCI combined with mindfulness therapy may improve sleep quality and reduce arousal by integrating neural feedback with psychological self-regulation[9]. EEG-based monitoring may help identify fatigue or suboptimal engagement during rehabilitation sessions; however, this application remains exploratory in stroke rehabilitation[19,42]. In addition, closed-loop systems can adjust the training load or feedback modality according to the patient’s brain state, sleep history, or fatigue level[8]. However, these hypotheses require prospective testing with repeated assessments of sleep, mood, fatigue, and adherence to rehabilitation during BCI-based interventions.

Closed-loop neuromodulation and individualized rehabilitation

One major potential benefit of the BCI is the opportunity for closed-loop personalized rehabilitation. Conventional rehabilitation is performed using fixed training protocols and standard exercises. In contrast, BCI systems may be able to provide feedback to the user based on brain activity signals in real time. Theoretically, when combined with wearable technology, sleep monitoring, heart rate variability, activity monitoring, and patient-reported outcomes, BCI have the potential to enable individualized rehabilitation.

For example, a future system could automatically adapt to the difficulty of training when fatigue is detected by EEG markers, offer relaxation-oriented feedback when hyperarousal is detected, or adapt the training intensity according to sleep quality from the previous night. Although such systems remain in the conceptual phase, they represent a promising direction for integrating motor, emotional, and sleep rehabilitation after a stroke. However, few stroke rehabilitation trials have implemented fully integrated multimodal closed-loop systems and their clinical feasibility remains to be established.

LIMITATIONS AND CHALLENGES

This minireview has a few limitations that should be considered when interpreting the results. The most critical problem is that direct evidence of BCI-based interventions for post-stroke emotional and sleep disorders remains scarce. To date, most investigations have concentrated on motor recovery, with emotional symptoms, sleep quality, fatigue, and quality of life treated as secondary or exploratory outcomes. Thus, the present findings should not be considered as an indication that BCI alone can improve post-stroke emotional or sleep disorders, but rather preliminary support for the broader rehabilitation potential of BCI.

Another challenge is the heterogeneity of BCI protocols. The existing studies differ in terms of signal acquisition methods, decoding strategies, feedback modalities, intervention doses, patient instructions, and control conditions. Such variability makes it difficult to determine the components that are most responsible for the clinical benefit. This issue is further complicated by the heterogeneity among stroke survivors. Differences in lesion characteristics, motor impairment, cognition, fatigue, communication ability, and baseline emotional or sleep statuses may influence BCI learning and training responses. In addition, some patients may have difficulty in achieving reliable BCI control despite adequate instruction, a phenomenon often described as BCI inefficiency or illiteracy.

Another implication of patient heterogeneity is the need for careful screening before BCI-based emotional or sleep rehabilitation. Patients with severe aphasia, marked cognitive impairment, attention deficits, severe fatigue, neglect, or poor task comprehension may have difficulty engaging in motor imagery, neurofeedback learning, or repeated closed-loop training. This is especially important in emotional rehabilitation because unsuccessful BCI control may increase frustration, reduce confidence, and undermine rehabilitation motivation. Thus, BCI inefficiency should not be viewed only as a technical classification problem, but also as a clinically meaningful barrier that may influence patient experience and adherence. Future studies should identify which patient subgroups are most likely to benefit from BCI-based emotional or sleep rehabilitation and which patients may require simplified interfaces, caregiver support, or alternative rehabilitation strategies.

FUTURE PERSPECTIVES

Future BCI research may progressively move beyond motor outcomes and should aim to assess post-stroke recovery in a multidimensional manner. Emotional symptoms, sleep quality, fatigue, participation, and quality of life are strongly associated with engagement in rehabilitation and recovery over the long term, and should be routinely included in BCI trials. BCI should not be considered a first-line therapy for post-stroke depression or insomnia. Future studies should investigate whether its use in BCI-based rehabilitation influences the scope of its effects through motor recovery, psychological engagement, neurofeedback learning, and/or the direct regulation of brain states.

A natural consequence of this is the categorization of upcoming trials according to the primary objective. A few trials may still be devoted to motor rehabilitation with non-motor measures as secondary endpoints, while others may evaluate integrated motor-psychological rehabilitation approaches, BCI, or neurofeedback protocols developed for emotional or sleep regulation. To enable a more reliable interpretation of these effects, future trials should use appropriate control conditions such as sham or non-contingent feedback, standard rehabilitation, or a similar intervention without BCI control. This would also help separate the BCI-specific effects from general therapy attention, motivation, and repeated rehabilitation training.

In addition to scientific validation, ethical and practical accessibility should also be considered before BCI can be widely integrated into stroke rehabilitation pathways. Current BCI systems require specialized equipment, trained personnel, repeated calibrations, and relatively long training sessions, which may limit their use in routine clinical settings and home-based rehabilitation. Cost, patient burden, caregiver involvement, data privacy, device maintenance, and long-term adherence are considered important implementation issues. Future development should therefore focus not only on improving decoding accuracy or feedback design but also on simplifying operations, reducing costs, supporting remote supervision, and ensuring equitable access for patients with different levels of disability and socioeconomic resources.

CONCLUSION

Recovery-based BCIs have advanced rapidly in post-stroke rehabilitation, with promising evidence from applications in motor rehabilitation using motor imagery, EEG-based decoding, FES, robotic feedback, and virtual reality. In addition, BCI-based strategies may have potential applications in post-stroke emotional and sleep disorders by enhancing active participation, neuroplasticity, self-regulation, and closed-loop feedback. These mechanisms are relevant because emotional disturbances and sleep symptoms can adversely affect engagement in rehabilitation, functional recovery, and quality of life.

The existing findings indicate that BCI may be a useful platform for integrated post-stroke rehabilitation of motor and non-motor functions. Specifically, BCI in conjunction with mindfulness practice, neurofeedback, FES, robotic assistance, or immersive feedback may offer new opportunities to address emotional symptoms, sleep quality, fatigue, and participation within an integrated rehabilitation framework. However, direct evidence specifically targeting post-stroke depression, anxiety, insomnia, and sleep quality remains limited, and the independent effects of BCI require further elucidation. With standardized outcome evaluations and well-designed clinical trials, BCI may gradually become a valuable component of comprehensive post-stroke rehabilitation.