The brain is a highly intricate network of anatomically interconnected neural components that interact continuously. A critical challenge in neuroscience is to understand how macroscale brain dynamics are shaped by the underlying microscale mechanisms (8). At the microscale, neuroplasticity relies on activity-dependent mechanisms that modulate the strength and efficiency of synaptic transmission. Key processes involved neurotransmitter release, calcium ion channels, N-methyl-D-aspartate (NMDA) receptors, α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptors, and intracellular signaling pathways (9). High-frequency activation of AMPAR induces Ca²⁺ influx into the postsynaptic neuron via NMDAR and L-type Ca²⁺ channels. This elevation in intracellular Ca²⁺ concentration resulting in the integration of AMPARs into the postsynaptic membrane, thereby improving postsynaptic responses. The aforementioned process of synaptic strengthening is called long-term potentiation (LTP), whereas the reduction in synaptic strength through AMPAR endocytosis is called long-term depression (LTD) (10). Abnormal changes in LTP and LTD lead to reduced synaptic efficiency, impairing the brain’s capacity to receive and process external information, ultimately manifests as cognitive impairments. Decades of neurobiological studies have demonstrated that neuromodulation strategies can improve cognitive function in PSCI by influencing cortical microstructural properties, including neurotransmitter levels, cytoarchitecture, dendritic cell type size, density and distribution, synaptic structure, and gene expression (8).

TMS enhanced synaptic density and active zone length, reduced synaptic cleft width, and upregulated the levels of PSD95, SYN, and brain-derived neurotrophic factor (BDNF) in the hippocampus (11). At the functional level, High-frequency repetitive TMS (HF-rTMS) enhanced LTP at CA3–CA1 synapses and synaptic plasticity by modulating the vascular endothelial growth factor (VEGF) and BDNF–NMDAR signaling pathways in the hippocampus, thereby facilitating spatial learning capacity in PSCI rats (12). Furthermore, rTMS has been shown to regulate the expression of genes related to glycinergic, glutamatergic, and GABAergic neurotransmission systems in cerebellum and brainstem, suggesting that it may modulate neuronal activity and synaptic plasticity by altering neurotransmitter levels (13). GABAergic activity plays a crucial role in shaping induced and evoked cortical oscillations, particularly in enhancing theta and gamma oscillations following a stroke (14). Wu et al. (15) suggested that intermittent theta-burst stimulation (iTBS) enhances hippocampus-dependent memory, potentially through augmented theta oscillations and changes in extracellular GABA and glutamate levels in the dorsal hippocampus. Similarly, the combination of transcranial direct current stimulation (tDCS) and donepezil has been shown to mitigate or delay cognitive impairment, enhance memory recall, and elevate acetylcholine levels in the cerebral cortex of stroke patients (16). VNS may promote norepinephrine release, thereby enhancing spatial and fear memory following cerebral ischemia (17).

Mesoscale mechanisms are processes that occur within neural circuits, which consist of connections between various types of neurons across brain regions. iTBS-induced cognitive improvement was related to the activation of the dorsolateral prefrontal cortex (DLPFC) and other distant regions, thereby improving assessment, decision-making capacities and cognitive control. In contrast, tDCS-induced cognitive improvement was associated with the activation of the frontopolar cortex, thereby enhancing valuation, motivation, and decisional substructures in PSCI patients (18). Furthermore, activation in the left superior temporal cortex, as measured by functional near-infrared spectroscopy (fNIRS), is also one of the mechanisms underlying tDCS treatment for PSCI (19).

Brain oscillations are rhythmic patterns of neural activity that emerges from interactions between intrinsic neuronal properties and the dynamics of interconnected neural networks (20). At the mesoscale, neuronal death in brain regions disrupts signal transmission within neural circuits, leading to abnormal structural or functional connectivity and neural oscillations. In a pilot trial, researchers found that theta-frequency rhythmic TMS targeting the parietal cortex improved performance on an auditory Working memory (WM) task by enhancing the oscillatory entrainment of frontoparietal theta rhythms (21). Song et al. (22) indicated that HF-rTMS reduces gamma oscillatory activity, enhances theta- and alpha-band functional connectivity within the DLPFC, and facilitates cognitive recovery in patients with PSCI. The application of transcranial alternating current stimulation (tACS), which involves the induction of a rhythmic oscillating flow of exogenous current associated with the endogenous brain oscillation models, can induce and affect endogenous brain oscillations that are frequency specific. This process is known as “neuronal entrainment” (23). Applying high-definition (HD)-tACS at alpha frequency to the unaffected hemisphere may enhance spatial attention in stroke patients by supporting the activity of unilateral alpha oscillations (24). Similarly, studies have shown that transcranial random noise stimulation (tRNS) regulates theta and gamma activity during the WM encoding phase and is more effective than tDCS in improving WM in healthy participants (25). Furthermore, Zheng et al. (26) demonstrated that 40 Hz light flicker alleviates the reduction of low gamma oscillations in the hippocampal CA1 region following cerebral ischemia and facilitates cognitive recovery. Mechanistically, it improves RGS12-regulated, CA3-CA1 presynaptic N-type calcium channel-dependent short-term synaptic plasticity and promotes postsynaptic LTP a model of cerebral ischemia. Overall, entrainment through rhythmic stimulation near or at the brain’s natural oscillations can facilitate more precise and personalized neuromodulation.

Evidence from preclinical and clinical neuroimaging studies indicates that cognitive impairment after stroke results from alterations in whole-brain networks rather than the loss of a single component (27). At the macroscale, neuromodulation facilitates whole-brain functional network reorganization, ultimately improving the overall function and behavior of stroke patients. Targeting specific cerebral regions with TMS in PSCI patients affects not only the directly stimulated areas but also distant brain regions through neural network interactions, resulting in lasting changes in inter-brain connectivity that underlie TMS-induced cognitive improvements. After rTMS treatment, patients with PSCI exhibited higher fractional amplitude of low-frequency fluctuation (fALFF) levels in the inferior frontal gyrus, superior temporal gyrus, and parahippocampal gyrus, and lower fALFF levels in the middle frontal gyrus, middle temporal gyrus, and fusiform gyrus. Additionally, rTMS enhanced functional connectivity (FC) between DLPFC and toprecuneus, marginal gyrus, inferior temporal gyrus, and middle and inferior frontal gyrus, while reducing FC between DLPFC and thalamus and middle temporal gyrus (28). Moreover, VNS is more effective in enhancing executive functions owing to elevated activity in the brainstem nuclei and nucleus of the solitary tract (NTS). Specifically, projections from the NTS to the locus coeruleus and dorsal raphe nucleus facilitate the circulation of monoaminergic neurotransmitters. Norepinephrine circulation may enhance functional connectivity between the prefrontal cortex (PFC), hippocampus, and amygdala, thereby promoting the formation of episodic memories, improving attention and alertness, and modulating emotional responses (29).

In this section, we reviewed neuromodulatory mechanisms at three scales: microscale (single neurons), mesoscale (neuronal populations), and macroscale (interregional interactions). Indeed, a comprehensive understanding of neuromodulatory mechanisms cannot be attained by examining any individual scale in isolation. From the perspective of systems neuroscience, stroke recovery is a complex, multiscale process emerging from the interplay among microscopic, mesoscopic, and macroscopic dynamics. To address this limitation, Shirinpour et al. (30) developed Neuron Modeling for TMS (NeMo-TMS), a multiscale modeling toolbox enabling simultaneous investigation of single-neuron responses to TMS across macro-, meso-, micro-, and subcellular scales. By integrating experimental findings into a unified computational framework, NeMo-TMS allows researchers to probe TMS mechanisms, thereby facilitating cross-scale understanding. Currently, this advanced technology has not yet been directly applied to stroke research. Further developments can be envisioned to use NeMo-TMS to validate the accuracy of neuronal membrane voltage and calcium dynamics in the neuromodulation of PSCI patients, thus facilitating a deeper understanding of PSCI’s physiological mechanisms and guiding the optimization of treatment parameters (Figure 2A).

To illustrate the efficacy of rTMS, we referred to a network meta-analysis with a large sample size that focused specifically on PSCI. This study evaluated 74 studies (n = 5478). The majority of these studies were randomized controlled trials (RCTs) with sham controls, although most involved small pilot samples (n < 50). The extant evidence suggests that rTMS can improve cognitive and daily life abilities of PSCI patients, with most studies targeting the DLPFC (40). DLPFC stimulation can enhance the deactivation of default mode network (DMN) nodes and facilitate cognitive processing under high demand, making it a preferred target for cognitive recovery (41). HF-rTMS and iTBS targeting the left DLPFC enhanced global cognition by increasing corticospinal excitability and instantly inducing theta oscillations, respectively, with HF-rTMS exerting a greater modulatory effect on attentional function than iTBS (28, 42, 43). LF-rTMS primarily targets DLPFC in the unaffected hemisphere to enhance cognitive function, and it reduces the number of magnetic stimulation pulses during treatment, thereby improving safety and patient tolerability (44). Mechanistically, LF-rTMS inhibits excitability in the contralateral cortex, thereby reducing its suppressive effect on the injured hemisphere and contributing to the restoration of interhemispheric inhibitory balance (45). The interhemispheric competition model may predominate in stroke patients with high structural reserve (the preservation of neural pathways and connectivity), whereas recovery in those with low structural reserve may be better explained by a hemispheric compensatory model (46). The hemispheric compensation model posits that when damage to the injured hemisphere is extensive and structural reserve is limited, rTMS excitation of the injured hemisphere is insufficient to support functional recovery. Instead, enhancing excitability in the healthy hemispheres is necessary to optimize treatment outcomes by expanding its compensatory functional capacity (47). Although the left DLPFC is the most commonly targeted region for TMS treatment of PSCI, the aforementioned bimodal balance-recovery model suggests that stimulation protocols should be adapted to the specific needs of individual patients. Furthermore, research has demonstrated that LF-rTMS targeting the contralesional primary motor cortex (M1) can enhance global cognitive function and visuospatial memory recall in stroke patients (48, 49). TMS targeting the posterior parietal cortex (PPC) has demonstrated therapeutic efficacy in stroke patients with unilateral spatial neglect (50). Using neuro-navigation to correct any positional or angulation errors of the TMS coil during the session will improve the accuracy of the target area (51).

In addition to the stimulation site, rTMS involves several key parameters, including frequency, intensity, and the number of pulses (52). In clinical settings, LF-rTMS is typically administered at 0.5 Hz or 1 Hz to reduce neuronal activity and cortical excitability, whereas HF-rTMS generally employs frequencies of 3 Hz, 5 Hz, 10 Hz, or 20 Hz to increase neuronal activity and cortical excitability. iTBS, a specific form of rTMS that replicates the brain’s innate theta rhythm, consists of delivering 3 pulses of 50 Hz bursts at 5 Hz (2s on and 8s off) for a total of 600 pulses, with each session lasting about 3 min (53). Using iTBS can instantly induce theta oscillations, which play a critical role in cognitive function and memory formation (43). Stimulus intensity varied between 70% and 120% of the resting motor threshold (RMT), with most studies selecting 80% RMT. The number of pulses varies significantly across studies (52). Different TMS parameters may influence outcomes in PSCI. However, these effects are not discussed in detail here, as they have been systematically examined in previous meta-analyses. In summary, a major challenge in the clinical practice of rTMS is the wide variation in the combination of parameters, and conducting orthogonally designed RCTs could help address this issue (Table 1).

TES administers a low-intensity electric current (1–4 mA) to the scalp, typically via two or more electrodes. According to the type of the applied current, TES can be categorized into tDCS, tACS, tRNS, transcranial pulsed current stimulation (tPCS), and temporal interference stimulation (TI) (63). The small size and portability of tDCS enhance its value as a neuromodulation tool (64). A clinical trial demonstrated that home-based tDCS with remote supervision can facilitate cognitive recovery in chronic stroke patients (65). When targeting deeper brain regions, such as the hippocampus (a region central to memory-related tasks) or the anterior cingulate cortex (a region involved in interference control), tDCS lacks sufficient penetration and focality to precisely target these areas without affecting the overlying cortex (66). However, the recently developed HD-tDCS can deliver a more focused current to the target region, partially addressing this limitation (67). Unlike TMS, tDCS does not directly trigger action potentials but modulates the resting membrane potential by regulating sodium-calcium channels and NMDARs, thereby altering the probability of neuronal discharge (68).

The mechanism of tACS differs fundamentally from that of tDCS in that no effect is observed on the average membrane potential. During each cycle, one electrode serves as the anode and the other as the cathode. In the subsequent cycle, the roles of the electrodes are reversed, ensuring that the average membrane potential remains unaltered (66). tACS targeting the supplementary motor area (SMA) could improve speech comprehension in stroke patients (69). Furthermore, a recent study developed cross-frequency (alpha-gamma) bifocal tACS (cf-tACS) by applying low-power sinusoidal currents at distinct frequencies to the primary visual cortex and the medio-temporal area. This study provides first evidence that a single session of cf-tACS can modify inter-areal coupling in healthy and stroke participants; however, long-term treatment is needed to further verify the behavioral effects (70). tRNS is a variant of tACS that transmits alternating currents at random intensities. In most tRNS studies, a frequency spectrum ranging from 0.1 Hz to 640 Hz (full spectrum) or from 101 Hz to 640 Hz (high-frequency stimulation) was employed (71, 72). It was proposed that tRNS enhances neural firing synchronization by amplifying subthreshold oscillatory activity, thereby reducing endogenous noise. This higher signal-to-noise ratio within the central nervous system, along with increased sensitivity in sensory processing, may contribute to augmented perception and cognitive performance (73). Of note, tRNS has not yet demonstrated therapeutic benefits for PSCI (Table 2).

tPCS is a recent advance in neuromodulation that converts direct current into pulses with programmable parameters, including pulse duration and inter-pulse interval (81). The primary mechanisms include charge summation by pulsed currents, which generates larger voltage gradients across neuronal ensembles, and phasic, subthreshold modulation within the region of interest induced by pulsed current stimulation (82). It has been demonstrated that tPCS exerts a frequency-specific effect on electroencephalographic band power, regulates functional connectivity, promotes interhemispheric coherence, and facilitates cortical plasticity (83). A case report provided preliminary evidence that tPCS targeting the bilateral DLPFC, parietal regions, and the precuneus improved memory, recognition, and other cognitive functions in patients with PSCI (84). TI represents another noninvasive approach for achieving effective deep brain neuromodulation. TI employs two high-frequency alternating currents of slightly different frequencies, delivered through two pairs of scalp electrodes. This approach produces a low-frequency modulated electric field (envelope modulation), enabling stimulation of deep structures while minimizing direct neural activation in the superficial cortex(F. 85). By oscillating the modulation envelope, TI can synchronize with ongoing neural activity through neural entrainment, similar to tACS (86). Violante et al. (87) applied sinusoidal currents of 2005 Hz and 2000 Hz to generate a 5 Hz envelope modulation targeting the left hippocampus, thereby achieving selective modulation of neural activity in the intended region. This intervention improved recall accuracy, suggesting that TI has the potential to enhance cognitive performance. An RCT will compare the efficacy of TI targeting the hippocampal region with tACS targeting the DLPFC in patients with PSCI, using standardized cognitive assessments and fNIRS to probe the underlying neural mechanisms (ChiCTR2400081207) (88).

VNS involves invasive VNS (iVNS) and non-invasive VNS, also called transcutaneous VNS (tVNS). tVNS is further subdivided into auricular (taVNS) and cervical (tcVNS) subtypes (89). taVNS offers a safe and well-tolerated alternative, engaging vagal pathways and projections with physiological responses similar to iVNS (90). MRI has shown that VNS causes localized alterations of blood flow within brain structures associated with cognition, including brainstem, hypothalamus, thalamus, amygdala, and hippocampus (91). In preclinical stroke studies, VNS has been demonstrated to decrease infarct volume and neurological deficits, thereby enhancing cognitive function by activating the cholinergic anti-inflammatory pathway, modulating blood-brain barrier integrity, and promoting angiogenesis and neurogenesis (92). An 8-week study by Chen et al. (93) showed that taVNS induced white matter remodeling in the bilateral DLPFC and enhanced global cognitive and executive function in a patient with chronic cerebral infarction. Despite its preliminary therapeutic potential, VNS for PSCI remains largely confined to preclinical studies and early-stage clinical investigations. PSCI is a chronic and heterogeneous condition that often requires long-term intervention to achieve functionally meaningful benefits. Whether improvements on short-term neuropsychological tests translate into real-world functional gains requires confirmation in large-scale clinical trials with long-term follow-up. Furthermore, the clinical outcomes of VNS are closely related to stimulation parameters, including stimulation site, on/off cycles, waveform, current amplitude, and frequency. Therefore, it is essential to elucidate these parameters to optimize VNS efficacy (Table 3).

PBM is a non-invasive, non-thermal form of neuromodulation in which cells and tissues are exposed to photons within specific wavelength ranges (96). Lasers and light-emitting diodes (LEDs) are the two most commonly used light sources in PBM. The precisely defined wavelengths of lasers are essential for experiments involving photosensitive compounds or mechanistic investigations. The high directionality and energy density of lasers enable precise targeting of the intended region, making them optimal for applications necessitating accurate, deep stimulation in medical practice. In contrast, LEDs offer a broader spectral range, making them more appropriate for general illumination and large-scale biological neuromodulation (97). Gao et al. (98) have demonstrated that laser-based PBM is more effective than LED-based PBM in enhancing cognitive function in patients. This finding warrants careful consideration, given that LEDs in clinical applications are safer, offer ease of home use, enable integration into wearable devices, and are low cost (99). Continuous wave and pulsed wave represent two principal modes of light energy output. Continuous wave is more appropriate for conditions that require long-term and steady irradiation, such as pain management, soft tissue inflammation, and wound healing. Pulsed wave exhibits remarkable efficacy in applications necessitating high peak power and expeditious response times (97). Tang et al. (100) employed psychological experiments and EEG to demonstrate that pulsed wave PBM produced greater cognitive enhancement than continuous wave PBM and modulated high-frequency neural oscillations. The primary mechanism of PBM involves cytochrome c oxidase (CCO), the principal photoreceptor for red-to-NIR light in neuronal mitochondria (101). Photons absorbed by CCO displace inhibitory nitric oxide (NO), resulting in increased mitochondrial membrane potential, enhanced oxygen consumption, and greater ATP production, which collectively reduce oxidative stress and neuroinflammation, ultimately mitigating neuronal damage following stroke (102).

Preclinical evidence indicates that PBM inhibits NOD-like receptor family pyrin domain-containing 3 (NLRP3) inflammasome activation and interleukin-1beta (IL-1β) expression in astrocytes of the hippocampal CA1 region, thereby promoting neural stem cell differentiation into neurons and supporting cognitive recovery (103). A clinical trial in stroke patients reported that, following the administration of transcranial photobiomodulation (tPBM) in combination with neuromuscular electrical stimulation, the experimental group showed a greater increase in mean Mini-Mental State Examination (MMSE) scores (24.85 ± 1.67 vs. 28.00 ± 1.73; p < 0.05) than the control group (26.28 ± 2.13 vs. 26.42 ± 1.27; p > 0.05) (104). Studies on healthy individuals provided a potential explanation for this phenomenon. Research has shown that tPBM targeting the right PFC can enhance visual WM capacity and elevate occipitoparietal contralateral delay activity (105). Additionally, tPBM modulates brain network topology, synchronization, and complexity, thereby facilitating more efficient neural information processing (106). Notably, direct clinical evidence supporting PBM for the treatment of PSCI remains limited. Future studies that examine light irradiation via the nasal and oral cavities, or the application of light-controlled nanomaterial drug-delivery systems to reduce light attenuation by the skull while enhancing safety, may further extend the clinical applications of PBM (Table 4).

TUS enables noninvasive, precise, stable, and targeted modulation of brain regions, making it an attractive option for both basic and clinical researchers (112). High-intensity focused ultrasound (HIFU) (I_SPPA > 200 W/cm²) is widely used for tumor thermocoagulation, intracerebral thrombus lysis, and the treatment of tremor associated with Parkinson’s disease and essential tremor. Low-intensity transcranial focused ultrasound (LIFU) (I_SPPA < 100 W/cm²) is regarded as a primary modality for non-invasive neuromodulation (113). This is because HIFU primarily relies on thermal effects, whereas LIFU mainly depends on mechanical bioeffects (114). Studies have shown that TUS can penetrate 5–6 cm into the cranium, enabling precise stimulation of deep brain lesions with millimeter-scale spatial resolution (115). To some extent, TUS overcomes the limitation that focal DBS currently requires the implantation of deep electrodes. Moreover, clinical neuromodulation requires the stimulation of pathological brain tissue, which often involves significant alterations in the brain’s normal conductivity. For electromagnetic techniques, accurately modeling these conductivity changes to achieve precise, individualized targeting remains nearly impossible. Conversely, ultrasound targeting is unaffected by such conductivity variations (116).

Multiple preclinical studies have indicated that TUS can restore cerebral blood supply and enhance the rate of vascular recanalization, attenuate inflammatory responses and apoptosis, and enhance neurotrophic factor expression, thereby conferring neuroprotection after stroke. However, whether TUS improves cognitive outcomes in stroke models remains largely unexplored (117–119). Only one clinical trial was identified that examined the effects of TUS in treating PSCI. Wang et al. (120) administered 20-minute daily sessions of TUS to the foreheads of stroke patients for six weeks. They reported that, compared to the sham stimulation group, patients in the TUS group exhibited significant improvements in executive function, naming, attention, language, and delayed recall, as well as higher levels of BDNF. Despite the positive results, the heterogeneity of testing conditions complicates the evaluation of these findings. Future research should test various parameter combinations and apply detailed cognitive function assessment tools to evaluate the effects of TUS for clinical practice.

BCIs are emerging communication and control systems that could revolutionize medicine by connecting the brain to external devices (121). A recent review indicated that BCI not only detects brain activity but also provides a means for assessing and training cognitive function in PSCI patients, facilitates intention expression, addresses cognitive and memory deficits, and enhances communication (122). Neurofeedback training (NFT) and motor imagery (MI) training are widely used approaches for enhancing cognitive function in BCI applications. T et al. (123)designed an EEG-BCI-based NFT game and evaluated its effectiveness in enhancing attention and memory among stroke patients, using both behavioral assessments and neurophysiological indices. The results showed that EEG-based attention scores in stroke patients improved by 4.29% to 32.18% from the first to the third session. Considering the interaction between sensorimotor and cognitive systems, a study examined the effects of MI-BCI training on upper limb function and attention in this population. The study found that MI-BCI can detect brain activity, classify and extract relevant information, decode motor intentions, and facilitate interneuronal communication. These mechanisms contribute to the rehabilitation of upper limb function and attention (124). Future research should balance the strengths and weaknesses of invasive and non-invasive BCI approaches to identify a more effective method for signal acquisition, while integrating AI, machine learning, and neural networks to develop bidirectional and high-performance BCIs (Table 5).

It has been hypothesized that TSS may enhance motor-related language functions, including action verb processing and speech articulation, by modulating activity within the sensorimotor network (130). Three studies performed by the same research group have evaluated the efficacy of TSS in patients with chronic post-stroke aphasia. They reported that anodal TSS improved verb naming performance, which was positively associated with cerebellar-cortical network connectivity. Additionally, patients showed greater accuracy in repeating the treated items (131–133).

Recent research suggests that DBS may rescue spatial memory in rats with cerebral ischemia by enhancing hippocampal synaptic activity (134). Although preclinical studies have indicated the potential efficacy of DBS in PSCI recovery, there is limited clinical evidence to support its use.

A meta-analysis involving 5,543 participants confirmed the therapeutic efficacy of MST in treating PSCI (135). Research indicated that MST can enhance depression, cognitive function, daily living activities, and neurological outcomes in stroke patients. The therapeutic mechanisms of MST may involve the evocation of positive emotions and relaxation, the influence of the autonomic and neuroendocrine systems, and the activation of the frontal, parietal, cerebellar, and limbic areas. Due to its susceptibility to personal preferences, emotional status, and cultural differences, MST may be better suited as an adjunct to other neuromodulation strategies (136).

Compared to traditional neuromodulation, genetic-based neuromodulation can simultaneously regulate multiple brain regions and produce more sustained effects (137). A recent study demonstrated that chemogenetic activation of neurons in the dorsal dentate gyrus alleviated anxiety, and depression in infarcted mice (138). Similarly, chemogenetic inhibition of ventral hippocampal CA1 pyramidal neurons reduced neurological injuries, anxiety, depression, and pain perception induced by cerebral ischemia through the cAMP response element-binding protein (CREB)-BDNF pathway (139). Furthermore, Chen et al. (140) found that early-stage chemogenetic activation of parvalbumin neurons ameliorated spatial working memory deficits induced by prefrontal ischemia in the T-maze task. Further studies revealed that parvalbumin neurons were suppressed during global ischemia but quickly recovered upon reperfusion. However, optogenetic stimulation of parvalbumin neurons induced GABAergic synaptic network activity remained significantly suppressed even one hour after reperfusion, potentially contributing to sensory and cognitive impairments observed after transient global ischemia (141).

Although chemogenetics and optogenetics have been demonstrated to precisely regulate neuronal excitability within specific brain regions, most studies have focused primarily on post-stroke motor dysfunction, while their direct application in models of PSCI remains relatively limited. Future research should focus on ensuring efficient receptor delivery and stable expression, improving ligand specificity, identifying the optimal timing and duration of genetic-based neuromodulation strategies, and clarifying the long-term consequences of chronic neural circuit modulation. Furthermore, emerging genetic-based neuromodulation strategies based on sound or magnetic stimulation may be more acceptable to patients due to their non-invasive or minimally invasive nature.