Efficacy and hematological markers of tDCS combined with aerobic exercise in improving executive function in patients with PSCI

Abstract

Objective

This study investigated the effects of combined transcranial direct current stimulation (tDCS) and aerobic exercise (AE) on executive function and hematological markers in poststroke cognitive impairment (PSCI) patients.

Methods

A total of 74 PSCI patients were randomly assigned to three groups: the tDCS + AE, tDCS, and AE groups. The tDCS group received 2.0 mA anodal stimulation over the left dorsolateral prefrontal cortex (DLPFC) for 30 min, five days per week for four weeks. The AE group performed moderate-intensity exercise (60–70% HRmax) for 30 min with sham tDCS. The combined group received both interventions concurrently. Executive function was assessed using C-EXIT25, the Stroop test, the 1-back task, and word recall tests. Emotional state was evaluated with the Hamilton Depression Rating Scale (HAMD) and the Hamilton Anxiety Rating Scale (HAMA). Serum biomarkers were analyzed pre- and postintervention. Statistical analyses included repeated measures ANOVA and Spearman correlation with Bonferroni correction.

Results

Executive function significantly improved in the combined group, with lower C-EXIT25 scores, faster Stroop congruent reaction times, and fewer word recall errors compared to the tDCS and AE groups. HAMD scores significantly decreased in the combined group. BDNF, NGF, DA, GABA, 5-HT, and ACh levels increased postintervention, whereas the level of Glu decreased in the combined group. An increase in BDNF correlated with faster Stroop response times, and an increase in NGF correlated with improvements in MMSE scores.

Conclusion

tDCS combined with AE significantly enhanced executive function in PSCI patients. These benefits were associated with increased neurotrophic factor levels, supporting a central-peripheral integrative rehabilitation approach for PSCI treatment.

Introduction

Poststroke cognitive impairment (PSCI) is a common and disabling consequence of stroke and represents a major challenge in long-term stroke rehabilitation. Among the various cognitive domains affected, executive dysfunction is consistently reported as the most prevalent and persistent deficit in PSCI patients [1–4]. Executive dysfunction encompasses impairments in attention control, working memory, inhibitory control, and cognitive flexibility, which are essential for goal-directed behavior and adaptive functioning in daily life. Deficits in these abilities are strongly associated with reduced independence, decreased participation in rehabilitation programs, and poorer functional outcomes after stroke [5–8]. Importantly, executive dysfunction often persists even when other cognitive domains show partial recovery and may remain evident for several years following stroke despite conventional rehabilitation approaches [5]. These characteristics highlight executive dysfunction as a critical therapeutic target in PSCI and underscore the need for effective, accessible, and sustainable intervention strategies.

Current treatment options for executive dysfunction in PSCI include both pharmacological and nonpharmacological approaches. Pharmacological therapies, such as cholinesterase inhibitors and N-methyl-D-aspartate receptor antagonists (e.g., memantine), have been widely investigated; however, clinical trials have reported inconsistent and generally modest effects on cognitive outcomes, particularly in the executive domain [3, 9, 10]. Moreover, the long-term use of these medications may be limited by adverse effects and poor tolerance, especially in older stroke survivors with multiple comorbidities. As a result, interest in nonpharmacological interventions that can be integrated into rehabilitation programs and delivered safely over extended periods is growing.

Transcranial direct current stimulation (tDCS) has emerged as a promising noninvasive neuromodulation technique for cognitive rehabilitation. By applying a weak direct current to targeted cortical regions, tDCS modulates neuronal membrane potentials and alters cortical excitability in a polarity-dependent manner. The dorsolateral prefrontal cortex (DLPFC) is widely recognized as a key hub of executive control and higher-order cognitive processing [10], making it a logical stimulation target for interventions aimed at improving executive function. Evidence-based guidelines have suggested that anodal tDCS over the left DLPFC may have positive effects on cognitive performance [11]. In stroke populations, several randomized controlled trials have demonstrated that tDCS can improve global cognitive measures, such as the Mini-Mental State Examination (MMSE) and the Montreal Cognitive Assessment (MoCA). However, findings regarding its specific effects on executive function in PSCI patients remain inconsistent. Some studies have reported improvements in overall cognition without corresponding gains in executive task performance, whereas others have reported limited or domain-specific benefits [12–14]. These mixed results suggest that tDCS alone may be insufficient to fully address the complex and heterogeneous nature of executive dysfunction following stroke.

Aerobic exercise (AE) is another well-established nonpharmacological intervention with documented benefits for cognitive function. A growing body of evidence indicates that moderate-intensity AE can enhance executive processes, including attention, working memory, and processing speed, in both healthy individuals and individuals with cognitive impairment [15, 16]. In patients with PSCI, AE has been shown to improve cognitive screening scores and executive task performance, particularly when delivered at moderate intensity and sufficient frequency [17, 18]. In addition to its cognitive benefits, AE offers well-known advantages for cardiovascular health, motor function, and overall physical conditioning, making it an attractive component of comprehensive stroke rehabilitation. Nevertheless, improvements in executive function achieved through AE alone are often modest, and considerable interindividual variability in treatment response has been reported [19]. These observations suggest that single-modality interventions may not adequately address the multifactorial mechanisms underlying executive dysfunction in PSCI.

Recent research has therefore shifted toward multimodal rehabilitation strategies that combine central and peripheral interventions to enhance cognitive recovery. From a theoretical perspective, combining tDCS with AE may produce synergistic effects by concurrently modulating cortical excitability and promoting systemic neuroplasticity. While tDCS directly influences neural activity within prefrontal executive networks, AE induces widespread physiological changes that support brain plasticity, potentially creating a more favorable neural environment for stimulation-induced effects [20, 21]. Preliminary studies in older adults and in patients with neurological disorders such as Parkinson’s disease and mild cognitive impairment suggest that the concurrent application of tDCS and AE yields greater improvements in executive function and related outcomes than either intervention alone [20, 22, 23]. Despite these encouraging findings, high-quality randomized controlled trials specifically examining the combined effects of tDCS and AE on executive dysfunction in PSCI patients are still lacking.

Given the high prevalence and long-term impact of executive dysfunction after stroke, as well as the limitations of existing single-modality interventions, there is a clear need to explore integrated rehabilitation approaches that target executive control more effectively. To investigate the synergistic effects of tDCS and AE on executive function, our team previously conducted a randomized controlled study [24]. A total of 60 healthy young participants were enrolled. The results revealed that compared with either tDCS or AE alone, the combination of tDCS stimulation and moderate-intensity AE significantly enhanced executive function in healthy young adults. This was manifested in improvements in inhibitory control (shorter reaction times in the incongruent condition of the word-color Stroop task), working memory (higher accuracy rates in the 2-back and 3-back tasks), and flexible switching ability during movement (shorter T test times). Moreover, the increase in blood BDNF concentrations in the combined group was more pronounced than that in the tDCS-alone or AE-alone groups. Additionally, the increase in blood BDNF levels across the three groups was negatively correlated with the shortened reaction times in the incongruent Stroop task. These findings laid the groundwork for our subsequent rehabilitation research on executive function in PSCI patients. Therefore, the present randomized controlled trial aimed to investigate the efficacy of combining anodal tDCS over the left DLPFC with moderate-intensity aerobic exercise in improving executive function in patients with PSCI. In addition to behavioral outcomes, we examined changes in circulating neuroplasticity-related biomarkers to provide insight into the potential mechanisms underlying the combined intervention. We hypothesized that the combined tDCS and AE intervention would lead to greater improvements in executive function than the intervention administered alone.

Methods

Research subjects

Stroke patients hospitalized in the affiliated mental health center of Jiangnan University were recruited to participate in this study, and the patients were informed of the possible risks before the trial.

Inclusion and exclusion criteria

The inclusion criteria were as follows: ① patients who met the diagnostic criteria for stroke, with the first occurrence of stroke in the frontal, parietal, temporal lobes, or basal ganglia; ② age ≥ 60 years, with an educational level of elementary school or higher; ③ disease duration ≥ 12 weeks; ④ right-handedness; ⑤ Ischemia Index Score ≥ 7; ⑥ No history of the following prior to stroke onset (as confirmed through detailed inquiries with the patient and their family members): a formal diagnosis of cognitive impairment, such as Alzheimer’s disease or vascular dementia; a previous neuropsychological assessment or treatment due to cognitive decline; significant impairment in activities of daily living (e.g., inability to independently manage finances, drive, or use complex tools); other preexisting psychiatric disorders, such as depression or anxiety; or long-term use of medications that may affect cognitive function (e.g., benzodiazepine sedatives or anticholinergic drugs); the patients also had to be fully or mostly independent in daily self-care activities prior to stroke onset (e.g., independently dressing, eating, and toileting). Poststroke cognitive impairment was screened using the MMSE based on the diagnostic criteria established by the Clinical Diagnosis and Research Center for Neurodegenerative and Memory Disorders at Xuanwu Hospital, Capital Medical University: for individuals with an elementary education (≤ 6 years), a score ≤ 20; for those with secondary education or higher (> 6 years), a score ≤ 24. Additionally, a score ≥ 15 on the 25-item Chinese Version of the Executive Interview (C-EXIT25) was used to identify executive dysfunction [25]; ⑦ no use of medications affecting cognitive function or heart rate prior to enrollment; ⑧ a willingness and ability to complete neuropsychological tests; ⑨ good visual, auditory, and verbal abilities, with no color blindness or color weakness; and ⑩ voluntary signing of the informed consent form by both the patient and their family members.

The exclusion criteria were as follows: ① unstable vital signs, progressive stroke, or recurrent stroke; ② the presence of metal implants in the skull or skin damage at the stimulation site; ③ the presence of epilepsy or other neurological or psychiatric disorders; ④ severe cardiopulmonary disease, renal failure, liver failure, or malignant tumors; ⑤ visual or auditory impairments; and ⑥ other conditions that prevent cooperation in completing the trial.

The withdrawal criteria were as follows: ① unwillingness to continue during the trial and voluntarily withdrawal from the trial; ② severe complications or worsening of the condition during the study, such as progressive or recurrent stroke, severe infection, or deep vein thrombosis of the lower limbs; ③ severe adverse reactions, such as severe headache or epileptic seizures; ④ patients or family members’ unwillingness to cooperate with the treatment plan; ⑤ patients or family members’ requests to voluntarily withdraw from the trial; and ⑥ patients who could not complete the trial because of various uncontrollable factors.

Sample size calculation and grouping

This study was a randomized controlled trial and was divided into three groups, namely, the combined (tDCS + AE) group, the tDCS group, and the AE group. C-EXIT25 was used as the primary efficacy indicator. According to the results of the pilot test (10 patients in each group), namely, the difference in C-EXIT25 scores, the average difference in the combined group was (-4.20 ± 2.29), the average difference in the tDCS group was (-1.80 ± 1.39), and the average difference in the AE group was (-1.90 ± 1.44). The effect size was 0.65. Assuming bilateral α = 0.05 and a power value of 0.95, a total sample size of n = 50 cases was calculated using GPower 3.1 software, that is, the sample size of n = 17 cases in each group was required. Considering the 20% dropout rate, at least 21 patients in each group were included as research subjects.

A total of 92 stroke patients (48 males and 44 females) were recruited to participate in the trial. Five patients (3 males and 2 females) who did not meet the inclusion criteria were excluded, and the remaining 87 patients (44 males and 43 females) were randomly divided using SPSS 26.0. Men were numbered 1 to 44 in order of enrollment. Initially, a starting point was set in the random number generator, and then a random number was generated for each participant using Rv.Uniform (0, 1) in Compute Variable. Afterward, a cutoff point was set in the “Visual Binning Method Based on Scanning Cases” to randomly divide the male patients into a combined (tDCS + AE) group, a tDCS group, and an AE group. Afterward, these operations were repeated to ensure that female patients were randomly assigned to the three experimental groups. Given the unique characteristics of aerobic exercise, it was not feasible to implement blinding for the study patients in this research. However, the assessors remained blinded to the allocation of the subjects, whereas both the patients and the personnel who delivered the intervention were not blinded.

During the experiment, owing to changes in the patient’s condition, 25 people were included in the combined group (13 males, 12 females), 25 were included in the tDCS group (12 males, 13 females), and 24 were included in the AE group (13 males, 11 females) (see Fig. 1).

Fig. 1.

PSCI patient enrollment flow chart

To minimize the interference of the experiment, all interventions were completed in the same treatment room with the same instrument, and the tDCS and AE interventions were completed by the same senior supervisor rehabilitation therapist who had received professional and standardized training. The trial protocol was explained to the patients and their families before the trial, and the patients were required to wear comfortable clothes and shoes to participate in the trial. After each trial, a questionnaire about tDCS was completed, covering side effects such as skin redness, mild headache, tingling, itching, burning, drowsiness, fatigue, inattention, and acute mood changes. If serious side effects occurred, the trial was terminated.

Ethical review

This study was approved by the Hospital Ethics Committee (Ethics WXMHCIRB2021LLky145) and was carried out in accordance with the Declaration of Helsinki. The trial was registered with the China Clinical Trial Registration Center (ChiCTR2100052745). All patients signed informed consent before being randomly assigned to each group.

Research methods

Research design

This study adopted a randomized controlled design and randomly divided the patients into three intervention groups to participate in an intervention 5 days a week (Monday to Friday) for 4 weeks. The three intervention groups were as follows: (1) the tDCS group, in which anodal tDCS was used only during rest; (2) the AE group, who performed AE under sham tDCS stimulation; and (3) the combined group, who performed AE under anodal tDCS stimulation. The patients were evaluated for behavioral characteristics, and relevant neurotrophic factors and neurotransmitters were measured one day before and 4 weeks after the intervention (Fig. 2). Following the completion of each assessment of all behavioral characteristics, participants were given a 2- to 5-minute rest period to mitigate the risk of participant fatigue, which could otherwise compromise their ability to cooperate with subsequent evaluation tasks. Additionally, blood biomarker sampling was scheduled for 7:00 AM to ensure that participants were in a fasting state, thereby optimizing the reliability of the metabolic and biochemical measurements.

Fig. 2.

Study protocol

Treatment and intervention methods

1. Basic treatment and routine rehabilitation treatment

In accordance with the clinical guidelines for stroke, a series of standardized treatment measures were implemented, including anti-platelet aggregation, improving blood circulation, nourishing nerves, regulating blood lipids and stabilizing plaques. For patients with coexisting basic diseases, corresponding treatment measures, such as antihypertensive therapy and blood sugar control, were administered.

The three groups of patients were administered routine rehabilitation treatment according to their specific functional disorders, which included mainly physical therapy, occupational therapy and cognitive function training. Physical therapy was based on the results of the patient’s rehabilitation assessment, and the specific content included good limb positioning, muscle strength training, joint range of motion training, stretching training, body control training, balance training, transfer training, standing training, gait training, etc., and was conducted once a day for 40 min. Occupational therapy was formulated by the therapist according to the patient’s functional condition, including rollers, wooden plugboards, and frosted boards, which were administered for 30 min each time, once a day. Cognitive function training included memory training, attention training, orientation training, visual perception training, spatial perception training, judgment and reasoning ability training, and executive ability training, which were administered for 30 min each time, once a day. Rehabilitation treatment was carried out 5 days a week (Monday to Friday) for 4 weeks. Each treatment was completed by a fixed therapist, who received standardized training and passed the assessment before the trial.

• (2)tDCS

Patients received tDCS in a quiet state. tDCS was provided by a battery-driven stimulator (YZB/Chuan 0185–2014; Sichuan Medical Device No. 20142210040; Sichuan Intelligent Electronics Industry Co., Ltd.) through a pair of saline-soaked sponge electrodes (7 cm × 5 cm) to stimulate the left DLPFC (F3), the main brain area for executive function, via anodal stimulation. The cathode electrode was placed in the contralateral supraorbital area. After the electrodes were placed, the patients were asked if they had tightness and discomfort so that the tightness of the band could be adjusted in a timely manner. During the use of tDCS, the stimulation current intensity was 2.0 mA and lasted 30 min, accelerating at the beginning of the stimulation and decelerating at the end of the stimulation and each stimulation lasting 30 s; tDCS was applied 5 consecutive days a week for 4 weeks. The placement of the electrodes for sham stimulation was the same as that of real stimulation tDCS. To simulate the slight tingling sensation caused by real stimulation and ensure the blinding of the electrical stimulation, the sham stimulation was set to a slow rise and fall current lasting 30 s at the beginning and end of each stimulation, but there was no current output when the patient performed the AE. Patients were blinded to the type of tDCS during the intervention (Fig. 3). After the trial, a tDCS experience questionnaire was administered to ask patients whether they felt any side effects of the stimulation, including mild discomfort or mild headache, neck pain, scalp pain, tingling, burning, and itching.

Fig. 3.

Testing program

• (3)AE

Owing to the limitation of limb movement in PSCI patients, AE was performed on a stationary power bicycle (model THERA-VIVAL, China). The seat height of the power bicycle was adjusted to an appropriate position according to the individual differences of the patients. The seat height of each patient was recorded and kept congruent in subsequent tests. The intensity of the moderate-intensity AE was individualized according to the individual’s maximum heart rate (HRmax) (HRmax = 220 - age). The moderate-intensity range was 60% to 70% of the HRmax [17]. The resistance of the power bicycle was adjusted so that the patient could reach the moderate-intensity target heart rate range. The exercise lasted for 40 min and included three stages: (1) 5 min of warm-up exercise, maintaining the heart rate between 55% and 60% of HRmax; (2) 30 min of peak exercise, maintaining the heart rate between 60% and 70% of HRmax; and (3) 5 min of cool-down exercise, maintaining the heart rate below 65% of HRmax. During exercise, the room temperature was kept constant at 24 to 26 °C, the patient’s upper body was kept as stable as possible, and a heart rate monitor (model DB18; Philips, Suzhou Erda Medical Equipment Co., Ltd.) was used to record the HR 5 consecutive days a week for 4 weeks (Fig. 3).

In cases where patients missed a scheduled treatment session because of temporary medical examination, mild physical discomfort (excluding severe adverse events), or other justifiable reasons, researchers arranged makeup sessions on weekends (Saturdays or Sundays). Data was retained for analysis if a participant completed at least 85% of the total prescribed treatment sessions (i.e., 17 or more sessions); otherwise, their data was excluded from the primary analysis.

Evaluation method

In accordance with the clinical characteristics of PSCI patients, the following specific evaluation methods were used: C-EXIT25 was used to evaluate overall executive function; the classic word-color Stroop task was used to evaluate the inhibitory control task of executive function; the 1-back task, word recall test, and word delayed recall test were used to evaluate the working memory task of executive function; the Trail making test (TMT) A and B (Chinese version) were used to evaluate the flexible switching task of executive function; the MMSE score was used to evaluate overall cognitive task execution; the Vascular Dementia Assessment Scale-Cognitive Section (VaDAS-cog) was used to evaluate overall cognitive status; the Fugl–Meyer score was used to evaluate motor performance on the hemiplegic side of the upper and lower limbs in the moving state; the modified Barthel scale was used to evaluate daily living ability; and the Hamilton Depression Scale (HAMD) and Hamilton Anxiety Scale (HAMA) were used to evaluate the emotional state of the patients.

1. Executive function assessment

① C-EXIT25 Correlational study of the Chinese version of the Executive Interview (C-EXIT25) to other cognitive measures in a psychogeriatric population in Hong Kong Chinese individuals. This scale comprises 25 items, each scored on a 0–2 point scale. The total score ranges from 0 to 50 points, with ≥ 15 points indicating abnormality. A higher score reflects more severe impairment in executive function.

② Word-color Stroop task The word-color Stroop test task was written using E-prime 3.0 software and run on a Lenovo Thinkpad X13 Ge laptop computer with a screen width of 28.5 cm and a height of 17.5 cm. The computer resolution when the program was running was set to 1920 × 1080 pixels, the refresh rate was 60 Hz, and the distance between the patient and the display screen ranged from 60 to 80 cm. The experiment used a module consisting of two types of task conditions: a congruent condition and an incongruent condition. The congruent condition trials consisted of one of three Chinese words for a color printed in the same color (red, yellow, and green), whereas the incongruent condition trials consisted of the same three Chinese words for a color printed in different colors (for example, the Chinese word for yellow was printed in the color green). Patients were required to respond to the corresponding keystrokes used in the writing color of the currently presented Chinese characters, ignoring the semantics of the presented Chinese characters. The “1”, “2”, and “3” keys on the keyboard were each used to represent a color: “1” represented red, “2” represented yellow, and “3” represented green. Patients completed the task with their healthy hand. In each trial, a fixation point was presented for 500 ms, followed by a Chinese character, which was presented until the patient responded. This block consisted of 120 conditional trials, including 60 congruent conditional trials and 60 incongruent conditional trials, and the presentation of the trials was random. The patient was asked to complete a practice module consisting of 10 trials. After the practice, the formal test began, and the accuracy of the patient’s congruent condition trials and incongruent condition trials and the reaction time of each trial were recorded as indicators of executive control inhibitory control (Fig. 4) [24].

Fig. 4.

Word‒color stroop task

③ 1-back task Consistent with the word-color Stroop task, the test was written using E-prime 3.0 software. The 1-back digital task was used to test working memory using a module consisting of two types of tasks: hit trials and correct rejection trials. The patient was asked to compare the current number with the number before it. The patient completed the task with their healthy hand and responded by pressing the “1” or “2” key. The “1” key indicated that the number matched the previous number, and the “2” key indicated that the number did not match the previous number. In each trial, the digital stimulus was presented for 800 ms, followed by a 2000 ms blank screen. The patient responded by pressing the key during the current stimulus presentation and the subsequent blank screen time. The 1-back task consisted of 120 conditional trials (60 hit trials and 60 correct rejection trials), and the presentation of the trials was random. The patient completed a practice module consisting of 10 trials. After the practice, the formal test began. The accuracy of the patient’s hit trials and correct rejection trials was recorded as an assessment indicator of executive function working memory (Fig. 5) [24].

Fig. 5.

1-back task

④ Word recall test and delayed word recall test The word recall test used was the word recall test and delayed word recall test in VaDAS-cog. The word recall test included 10 words. After the patient read them once, he or she was asked to remember as many words as possible and repeat them back to the experimenter. This process was repeated three times. The total score was the average number of words that could not be correctly repeated in the three tests, that is, the number of errors. The delayed word recall test used the same 10 words as in the word recall test. The number of words recalled after 5 min represented delayed memory. A 2-minute recall time was used, and the number of words that could not be correctly recalled was recorded. The fewer the number of errors in word recall and delayed word recall test, the better the participant’s memory ability [26].

⑤ TMT-A/B test: This test included two parts: A and B. The time taken to complete the test was recorded with a stopwatch and reflected the participant’s task switching ability in terms of executive function. The shorter the time taken, the better the reaction sensitivity [27].

• (2)Assessment of overall cognitive function

① MMSE score: This scale contains 30 items, with 1 point for each correct item and 0 points for each incorrect item, with a total score of 0 to 30 points. The score is interpreted on the basis of the participant’s level of education. Cognitive dysfunction is assessed as ≤ 20 points for those with a primary school education level (≤ 6 years) and ≤ 24 points for those a secondary school or above education level (> 6 years) [28].

② VaDAS-cog score: This scale contains 12 items and is often used to evaluate cognitive dysfunction caused by vascular diseases. It includes a word recall test (10 points), a delayed word recall test (10 points), instructions (5 points), structured exercises (5 points), naming objects or fingers (10 points), intentional exercises (5 points), an orientation test (8 points), a word recognition test (20 points), recall test instructions (5 points), verbal comprehension ability (5 points), word finding difficulty (5 points), and oral verbal expression ability (5 points). The higher the score is, the longer the time, and the more errors there are, the more severe the degree of cognitive decline [26].

• (3)Assessment of motor and daily living ability

① Fugl–Meyer score: This scale contains 50 items, with 33 items, each with a maximum score of 2 points and a total possible score of 66 points, allocated to the upper limbs and 17 items, each with a maximum score of 2 points and a total possible score of 34 points, allocated to the lower limbs. The total score for the upper and lower limbs is 100 points and is interpreted as follows: <50 points indicates severe motor impairment, 50 to 84 points indicates obvious motor impairment, 85 to 95 points indicates moderate motor impairment, and 96 to 99 points indicates mild motor impairment [29].

② Modified Barthel Scale: This scale contains 11 items, each of which has 5 options. Items 2, 3, and 10 are scored as 0, 1, 3, 4, and 5 points; Items 1, 4, 5, 6, 7, and 11 are scored as 0, 2, 5, 8, and 10 points; and Items 8 and 9 are scored as 0, 3, 8, 12, and 15 points. The total score ranges between 0 and 100. The scale assesses the patient’s ability to complete activities of daily living and determines whether the patient needs assistance with daily living activities. A score below 20 indicates complete dependency, 20 to 40 points indicates a great need for assistance, 40 to 60 points indicates a need for assistance, and a score above 60 points indicates that the patient is capable of basic self-care [30].

• (4)Emotional state assessment

① HAMD score: The scale contains 24 items. Most items use a 5-level scoring method, ranging from 0 to 4 points (none, mild, moderate, severe, and extremely severe, respectively). A few items use a scoring method that ranges from 0 to 2 points, corresponding to none, mild to moderate, and severe, respectively. The scores of all the items are added to obtain the total score, and a total score of < 8 is classified as no depressive symptoms [31].

② HAMA score: The scale contains 14 items. All the items are scored on a 5-point scale ranging from 0 to 4 points (no, mild, moderate, severe and extremely severe, respectively) [32].

Detection of related neurotrophic factors and neurotransmitters

Serum levels of BDNF, NGF, DA, GABA, Glu, 5-HT and ACh were determined using enzyme-linked immunosorbent assay (ELISA) kits provided by Lapuda Biotechnology Co., Ltd, Nanjing, China. Serum samples were diluted as recommended by the manufacturer, and the assay was performed in 96-well plates. The assays were carried out following the manufacturer’s recommended protocol, and the absorbance was measured using a microplate reader set at the appropriate wavelength. The lower detection limits (LODs) for each cytokine were 0.156 ng/mL for BDNF, 15.600 pg/mL for NGF, 0.947 pg/mL for DA, 24.69 pg/mL for GABA, 1.201 ng/mL for Glu, 1.103 ng/mL for 5-HT, and 0.984 pg/mL for AChE. Values lower than the LOD were reported as half of the LOD.

Statistical analysis methods

1. Behavioral statistical analyses

SPSS v.26.0 software (IBM, USA) was used for the statistical analyses. The patient demographic data were subjected to a normality test and variance homogeneity test. The data that were normally distributed are expressed as the mean ± standard deviation (mean ± standard deviation), and those that were not normally distributed were tested with a paired Wilcoxon rank sum test and are expressed as the median ± quartile (M ± QID). The count data are expressed as specific values. One-way analysis of variance was used for the continuous variables, and the chi-square (χ²) test was used for the categorical variables to determine differences among the three groups.

For the analysis of C-EXIT25 scores before and 4 weeks after the intervention, the word-color Stroop task accuracy/reaction time, 1-back accuracy, word recall test error number, word delayed recall error number, TMT-A/B duration, MMSE scores, VaDAS-cog scores, Fugl–Meyer scores, modified Barthel scores, HAMA scores, and HAMD scores, a 3 × 2 repeated measures ANOVA was used, in which the between-group factors were 3 groups (tDCS + AE vs. tDCS vs. AE) and the within-group factors were 2 time points (before the intervention vs. after the intervention).

For results that did not meet the Mauchlys sphericity test, the Greenhouse–Geisser method was used to correct the degrees of freedom. If there was an interaction effect, the Bonferroni method was used for post hoc comparisons, and the F test effect size was expressed as η².

Considering practicability in clinical practice, a Spearman correlation analysis was performed between the clinical assessment scale, demographic data and baseline C-EXIT25 scores, and p < 0.05 was considered to indicate statistical significance. Afterward, the C-EXIT25 score after the intervention was used as the dependent variable, and the variables related to the C-EXIT25 score at baseline and the grouping status were used as independent variables for multivariate linear regression analysis to predict the effects of the intervention on the improvement of executive function in PSCI patients.

1. Statistical analysis of neurotrophic factors and neurotransmitters

For the analysis of the concentrations of BDNF, NGF, DA, GABA, Glu, 5-HT, and ACh in the blood before and after 4 weeks of intervention, a 3 × 2 repeated measures analysis of variance was used, in which the intergroup factors were 3 groups (tDCS + AE vs. tDCS vs. AE), and the intragroup factors were 2 time points (before the intervention vs. 4 weeks after the intervention).

For the results that did not meet the Mauchlys sphericity test, the Greenhouse–Geisser method was used to correct the degrees of freedom. If there was an interaction effect, the Bonferroni method was used for post hoc comparisons. When the test level was α = 0.05, the significance level of the bilateral test was set at p < 0.05, and the F test effect size was expressed as η².

Spearman correlation was used to analyze the correlation between the differences in neurotrophic factors and neurotransmitters in the blood before and after the intervention and the differences before and after the behavioral intervention. The reference standards for correlation were as follows: <0.2 (no correlation), 0.2 ~ 0.4 (weak correlation), 0.4 ~ 0.7 (moderate correlation), and > 0.7 (strong correlation). Considering the multiple pairwise comparisons (7 × 21), Bonferroni correction (0.05/147) was used, and p < 0.0004 indicated a significant correlation.

Results

Behavioral analysis before and after the intervention

Demographic data and adverse reactions

A total of 74 patients were treated in this study, including 25 in the combined group, 25 in the tDCS group, and 24 in the AE group. The analysis revealed that there were no significant differences in age, sex, stroke type, stroke site, disease course, years of education, or BMI among the three groups (p > 0.05) (Table 1). In terms of adverse reactions, the main types of feedback from the questionnaire survey were mild tingling (100%, 100%), itching (96.00%, 96.00%), burning (80.00%, 80.00%), drowsiness (4.00%, 0%), skin redness (80.00%, 80.00%), and skin irritation (80.00%, 80.00%) in the combined group and in the tDCS group, but no patient discontinued the study due to adverse reactions, and no feedback regarding serious side effects was received (Tables 1 and 2).

Table 1.

Demographic data of the patients in the three groups (n = 74)

Demographic and general information	tDCS + AE group	tDCS group	AE group	F/H/χ2	P value
Age (years)	67.76 ± 7.29	68.92 ± 5.86	69.71 ± 5.63	0.592	0.556
Sex (male/female)	13/12	12/13	13/11	0.193	0.908
Stroke type (infarction/hemorrhage)	21/4	20/5	20/4	0.158	0.924
Stroke location (frontotemporal/basal ganglia)	13/12	11/14	11/13	0.351	0.838
Years of education (years)	9(5,12)	9(5.5,12)	9(7.25,11.25)	0.153	0.926
BMI (kg/m2)	24.97 ± 3.35	24.49 ± 3.68	24.23 ± 3.56	0.277	0.759

Table 2.

Adverse reactions of the patients in the three groups (n = 74)

Adverse reactions	tDCS + AE group	tDCS group	AE group
Headache	0	0	0
Neck pain	0	0	0
Tingling	25	25	0
Itching	24	24	0
Burn	20	20	0
Drowsiness	1	0	0
Metallic/rusty taste	0	0	0
Fatigue	0	0	0
Poor concentration	0	0	0
Acute mood changes	0	0	0
Skin redness	20	20	0
Skin irritation	20	20	0

Behavioral results before and after the intervention

1. Executive function assessment

In the executive function assessment, after the intervention, the C-EXIT25 scores, Stroop congruent accuracy, Stroop incongruent accuracy, Stroop congruent reaction time, Stroop incongruent reaction time and word recall incorrect response number of the three groups of patients significantly improved (p < 0.05). However, for the 1-back hit accuracy and word delayed recall incorrect response number, only the combined group and the tDCS group showed significant improvement (p < 0.05), and for the TMT-A duration, only the combined group showed significant improvement (p < 0.05). The time × group interaction effect was significant for the C-EXIT25 score, Stroop congruent reaction time, Stroop incongruent reaction time, 1-back hit accuracy, and word recall incorrect response number (p < 0.05). After the intervention, the C-EXIT25 score of the combined group was significantly lower than that of the tDCS group (p = 0.043) and the AE group (p = 0.004); the Stroop congruent reaction time was significantly lower than that of the tDCS group (p = 0.025) and the AE group (p = 0.048); and the word recall incorrect response number was significantly lower than that of the tDCS group (p = 0.014) and the AE group (p = 0.000). These findings reveal that the improvements in the C-EXIT25 score, Stroop congruent reaction time, and word recall incorrect response number in the combined group were more significant than those in the tDCS group and the AE group, as shown in Table 3; Fig. 6.

Table 3.

Analysis of the executive function assessment results in the three groups of patients (n = 74)

Executive Function	Group	Before the intervention	After the intervention	F1,71	p	η2p
C-EXIT25	tDCS + AE group	24.08 ± 2.31	19.64 ± 2.74	70.272	<0.001	0.497
	tDCS group	24.24 ± 2.57	21.68 ± 2.50	23.361	<0.001	0.248
	AE group	24.00 ± 2.28	22.38 ± 3.33	9.036	0.004	0.113
	F 2,71	0.064	6.053
	p	0.938	0.004
	η 2 p	0.002	0.146
Time (F2,71, p, η2p)	87.181	<0.001	0.551
Group (F2,71, p, η2p)	2.384	0.099	0.062
Time×Group (F2,71, p, η2p)	7.214	0.001	0.169
Stroop Con.ACC	tDCS + AE group	0.97 ± 0.06	0.99 ± 0.02	4.382	0.040	0.058
	tDCS group	0.95 ± 0.07	0.99 ± 0.02	12.171	0.001	0.146
	AE group	0.95 ± 0.07	0.99 ± 0.03	7.931	0.006	0.100
	F 2,71	0.573	0.377
	p	0.566	0.687
	η 2 p	0.016	0.011
Time (F2,71, p, η2p)	23.509	<0.001	0.249
Group (F2,71, p, η2p)	0.597	0.553	0.017
Time×Group (F2,71, p, η2p)	0.489	0.615	0.014
Stroop InCon.ACC	tDCS + AE group	0.82 ± 0.15	0.95 ± 0.09	27.814	<0.001	0.281
	tDCS group	0.81 ± 0.13	0.88 ± 0.09	8.202	0.005	0.104
	AE group	0.81 ± 0.12	0.87 ± 0.10	6.316	0.014	0.082
	F 2,71	0.072	4.916
	p	0.930	0.010
	η 2 p	0.002	0.122
Time (F2,71, p, η2p)	37.660	<0.001	0.347
Group (F2,71, p, η2p)	1.419	0.249	0.038
Time×Group (F2,71, p, η2p)	2.186	0.120	0.058
Stroop Con.RT (ms)	tDCS + AE group	4388.86 ± 619.34	2864.45 ± 849.03	263.568	<0.001	0.788
	tDCS group	4282.23 ± 553.64	3495.83 ± 629.82	70.141	<0.001	0.497
	AE group	4286.04 ± 746.07	3444.41 ± 957.86	77.126	<0.001	0.521
	F 2,71	0.220	4.530
	p	0.803	0.014
	η 2 p	0.006	0.113
Time (F2,71, p, η2p)	370.570	<0.001	0.839
Group (F2,71, p, η2p)	1.071	0.348	0.029
Time×Group (F2,71, p, η2p)	19.066	<0.001	0.349
Stroop InCon.RT (ms)	tDCS + AE group	6506.02 ± 624.89	5457.22 ± 850.78	287.082	<0.001	0.802
	tDCS group	6443.26 ± 599.01	5498.96 ± 708.36	232.727	<0.001	0.766
	AE group	6557.27 ± 967.20	5748.15 ± 1143.23	164.028	<0.001	0.698
	F 2,71	0.144	0.719
	p	0.866	0.491
	η 2 p	0.004	0.020
Time (F2,71, p, η2p)	673.776	<0.001	0.383
Group (F2,71, p, η2p)	0.375	0.688	0.010
Time×Group (F2,71, p, η2p)	3.686	0.030	0.094
Stroop interference effect (ms)	tDCS + AE group	2117.17 ± 657.44	-747.28 ± 1243.41	303.904	<0.001	0.811
	tDCS group	2161.03 ± 488.99	-1334.80 ± 947.25	452.644	<0.001	0.864
	AE group	2271.22 ± 814.08	-1173.19 ± 1301.69	421.848	<0.001	0.856
	F 2,71	0.348	1.672
	p	0.707	0.195
	η 2 p	0.010	0.045
Time (F2,71, p, η2p)	1170.608	<0.001	0.943
Group (F2,71, p, η2p)	0.625	0.538	0.017
Time×Group (F2,71, p, η2p)	4.530	0.014	0.113
1-back Hit.ACC	tDCS + AE group	0.72 ± 0.10	0.81 ± 0.07	35.143	<0.001	0.331
	tDCS group	0.70 ± 0.13	0.76 ± 0.10	13.752	<0.001	0.162
	AE group	0.73 ± 0.12	0.76 ± 0.10	3.581	0.063	0.048
	F 2,71	0.461	2.033
	p	0.633	0.138
	η 2 p	0.013	0.054
Time (F2,71, p, η2p)	43.994	<0.001	0.383
Group (F2,71, p, η2p)	0.672	0.524	0.019
Time×Group (F2,71, p, η2p)	3.934	0.024	0.100
1-back Correct rejection.ACC	tDCS + AE group	0.74 ± 0.11	0.73 ± 0.07	0.192	0.662	0.003
	tDCS group	0.72 ± 0.08	0.73 ± 0.05	0.057	0.812	0.001
	AE group	0.72 ± 0.14	0.71 ± 0.14	0.280	0.598	0.004
	F 2,71	0.213	0.391
	p	0.809	0.678
	η 2 p	0.006	0.011
Time (F2,71, p, η2p)	0.180	0.673	0.003
Group (F2,71, p, η2p)	0.318	0.728	0.009
Time×Group (F2,71, p, η2p)	0.178	0.838	0.005
1-back conflict effect size	tDCS + AE group	0.025 ± 0.127	-0.074 ± 0.097	19.282	<0.001	0.214
	tDCS group	0.024 ± 0.127	-0.027 ± 0.113	5.256	0.025	0.069
	AE group	-0.011 ± 0.136	-0.051 ± 0.117	3.048	0.085	0.041
	F 2,71	0.632	1.167
	p	0.535	0.317
	η 2 p	0.017	0.032
Time (F2,71, p, η2p)	23.561	<0.001	0.249
Group (F2,71, p, η2p)	0.555	0.576	0.015
Time×Group (F2,71, p, η2p)	1.890	0.159	0.051
Word recall incorrect response number	tDCS + AE group	5.19 ± 1.06	2.65 ± 1.48	134.896	<0.001	0.655
	tDCS group	5.32 ± 1.29	3.96 ± 1.85	38.915	<0.001	0.354
	AE group	5.65 ± 1.32	4.97 ± 1.37	9.340	<0.001	0.116
	F 2,71	0.922	13.147
	p	0.402	<0.001
	η 2 p	0.025	0.270
Time (F2,71, p, η2p)	144.597	<0.001	0.671
Group (F2,71, p, η2p)	6.904	0.002	0.163
Time×Group (F2,71, p, η2p)	18.63	<0.001	0.338
Word delayed recall incorrect response number	tDCS + AE group	5.08 ± 1.82	4.52 ± 1.38	9.919	0.002	0.123
	tDCS group	5.36 ± 1.68	4.76 ± 1.69	11.387	0.001	0.138
	AE group	5.33 ± 1.93	5.13 ± 1.78	1.318	0.255	0.018
	F 2,71	0.181	0.859
	p	0.835	0.428
	η 2 p	0.005	0.024
Time (F2,71, p, η2p)	19.469	<0.001	0.215
Group (F2,71, p, η2p)	0.426	0.662	0.012
Time×Group (F2,71, p, η2p)	1.430	0.246	0.039
TMT-A Duration (s)	tDCS + AE group	96.22 ± 19.65	93.03 ± 21.91	5.287	0.024	0.069
	tDCS group	97.48 ± 13.71	95.30 ± 12.97	2.490	0.119	0.034
	AE group	99.61 ± 11.68	98.40 ± 13.05	0.717	0.400	0.010
	F 2,71	0.299	0.649
	p	0.742	0.526
	η 2 p	0.008	0.018
Time (F2,71, p, η2p)	7.392	0.008	0.094
Group (F2,71, p, η2p)	0.485	0.617	0.013
Time×Group (F2,71, p, η2p)	0.504	0.606	0.014
TMT-B Duration (s)	tDCS + AE group	250.98 ± 30.29	250.86 ± 26.62	0.004	0.951	0.000
	tDCS group	251.06 ± 21.07	250.84 ± 18.17	0.014	0.908	0.000
	AE group	245.91 ± 20.69	246.91 ± 18.74	0.267	0.607	0.004
	F 2,71	0.354	0.270
	p	0.703	0.764
	η 2 p	0.010	0.008
Time (F2,71, p, η2p)	0.040	0.842	0.001
Group (F2,71, p, η2p)	0.326	0.723	0.009
Time×Group (F2,71, p, η2p)	0.124	0.883	0.003

Fig. 6.

Time (before the intervention vs. after the intervention) × group (tDCS + AE vs. tDCS vs. AE) interaction results of the executive function assessment results in the three groups before and after the intervention. A: C-EXIT25 score; B: Stroop congruent accuracy; C: Stroop incongruent accuracy; D: Stroop congruent reaction time; E: Stroop incongruent reaction time; F: Stroop interference effect time; G: 1-back hit accuracy; H: 1-back correct rejection accuracy; I: 1-back conflict effect size; J: word recall incorrect response number; K: word delayed recall incorrect response number; L: TMT-A duration; and M: TMT-B duration (****p < 0.0001, 0.0001< *** p < 0.001, 0.001 < **p < 0.01, 0.01 < * p < 0.05)

(2) Overall cognitive function assessments

In the overall cognitive function assessments, after the intervention, the MMSE and VaDAS-cog scores of the three groups significantly improved (p < 0.05), but the interaction effect of the MMSE score and the VaDAS-cog score time × group was not significant (p > 0.05), as shown in Table 4; Fig. 7.

Table 4.

Analysis of the overall cognitive function assessment results in the three groups of patients (n = 74)

Overall cognitive function assessment	Group	Before the intervention	After the intervention	F1,71	p	η2p
MMSE	tDCS + AE group	20.04 ± 2.21	22.40 ± 2.58	126.495	<0.001	0.640
	tDCS group	19.24 ± 1.76	21.00 ± 1.76	70.352	<0.001	0.498
	AE group	20.46 ± 1.91	21.38 ± 1.84	18.321	<0.001	0.205
	F 2,71	2.430	2.986
	p	0.095	0.057
	η 2 p	0.064	0.078
Time (F2,71, p, η2p	189.420	<0.001	0.727
Group (F2,71,p,η2p)	2.087	0.132	0.056
Time×Group (F2,71, p, η2p)	11.673	<0.001	0.247
VaDAS-cog	tDCS + AE group	23.68 ± 3.70	20.16 ± 3.50	28.066	<0.001	0.283
	tDCS group	24.33 ± 4.92	21.05 ± 4.38	24.369	<0.001	0.256
	AE group	23.88 ± 4.49	22.14 ± 3.55	6.555	0.013	0.085
	F 2,71	0.145	1.632
	p	0.865	0.203
	η 2 p	0.004	0.044
Time (F2,71, p, η2p)	54.264	<0.001	0.433
Group (F2,71,p,η2p)	0.540	0.585	0.015	54.264
Time×Group (F2,71, p, η2p)	2.066	0.134	0.055

Fig. 7.

Time (before the intervention vs. after the intervention) × group (tDCS + AE vs. tDCS vs. AE) interaction results of the overall cognitive function assessment results in the three groups before and after the intervention. A: MMSE score; B: VaDAS-cog score (****p < 0.0001, 0.0001< *** p < 0.001, 0.001 < **p < 0.01, 0.01 < * p < 0.05)

• (3)Assessment of exercise and daily living abilities

In the assessment of exercise and daily living abilities, after the intervention, the Fugl–Meyer and modified Barthel scores of the three groups of patients significantly improved (p < 0.05), but the time × group interaction was not significant (p > 0.05), which indicated that there was no significant difference in the improvement in the Fugl–Meyer and modified Barthel scores among the three groups (see Table 5; Fig. 8).

Table 5.

Analysis of the exercise and daily living ability assessment results in the three groups of patients (n = 74)

Exercise and daily living abilities	Group	Before the intervention	After the intervention	F1,71	p	η2p
Fugl–Meyer	tDCS + AE group	63.12 ± 16.14	67.48 ± 15.09	171.490	<0.001	0.707
	tDCS group	61.48 ± 16.72	65.76 ± 16.53	165.254	<0.001	0.699
	AE group	64.12 ± 17.13	67.58 ± 16.76	103.579	<0.001	0.593
	F2,71	0.158	0.100
	p	0.854	0.905
	η 2 p	0.004	0.003
Time (F2,71, p, η2p)	434.115	<0.001	0.859
Group (F2,71,p,η2p)	0.124	0.883	0.003
Time×Group (F2,71, p, η2p)	2.187	0.120	0.058
Modified Barthel	tDCS + AE group	59.52 ± 16.88	62.56 ± 15.97	35.687	<0.001	0.335
	tDCS group	58.44 ± 13.12	61.60 ± 13.11	38.560	<0.001	0.352
	AE group	60.63 ± 16.77	62.79 ± 16.32	17.403	<0.001	0.197
	F2,71	0.119	0.043
	p	0.888	0.958
	η 2 p	0.003	0.001
Time (F2,71, p, η2p)	88.871	<0.001	0.556
Group (F2,71,p,η2p)	0.075	0.928	0.002
Time×Group (F2,71, p, η2p)	1.104	0.337	0.030

Fig. 8.

Time (before the intervention vs. after the intervention) × group (tDCS + AE vs. tDCS vs. AE) interaction results of the exercise and daily living abilities assessments in the three groups before and after the intervention. A: Fugl–Meyer score; B: modified Barthel score (****p < 0.0001, 0.0001< *** p < 0.001, 0.001 < **p < 0.01, 0.01 < * p < 0.05)

(4) Emotional state assessments

In terms of the emotional state assessments, after the intervention, the HAMD and HAMA scores of the combined group and the tDCS group significantly improved (p < 0.05), and the time × group interaction of the HAMD and HAMA scores was significant (p < 0.05), but the HAMD scores of the combined group were significantly lower than those of the tDCS group (p < 0.001) and the AE group (p < 0.001). These findings reveal that the improvements in the HAMD scores in the combined group were more significant than those in the tDCS group and the AE group (Table 6; Fig. 9).

Table 6.

Analysis of the emotional state assessment results in the three groups of patients (n = 74)

Emotional state	Group	Before the intervention	After the intervention	F1,71	p	η2p
HAMD	tDCS + AE group	6.96 ± 1.46	4.72 ± 1.02	96.016	<0.001	0.575
	tDCS group	6.68 ± 1.38	6.16 ± 1.25	5.174	0.026	0.068
	AE group	7.13 ± 1.42	6.92 ± 1.47	0.797	0.375	0.011
	F 2,71	0.618	19.383
	p	0.542	0.000
	η 2 p	0.017	0.353
Time (F2,71,p,η2)	55.432	<0.001	0.438
Group (F2,71,p,η2p)	5.804	0.005	0.141
Time×Group (F2,71, p, η2p)	22.673	<0.001	0.390
HAMA	tDCS + AE group	6.44 ± 1.23	5.60 ± 1.38	21.510	<0.001	0.233
	tDCS group	6.44 ± 1.50	5.96 ± 1.14	7.024	0.010	0.090
	AE group	6.25 ± 1.26	6.12 ± 1.19	0.457	0.501	0.006
	F 2,71	0.164	1.150
	p	0.849	0.322
	η 2 p	0.005	0.031
Time (F2,71, p, η2p)	20.927	<0.001	0.228
Group (F2,71,p,η2p)	0.172	0.842	0.005
Time×Group (F2,71, p, η2p)	3.818	0.027	0.097

Fig. 9.

Time (before the intervention vs. after the intervention) × group (tDCS + AE vs. tDCS vs. AE) interaction results for the emotional state assessment results in the three groups before and 4 weeks after the intervention. A: HAMD score; B: HAMA score (****p < 0.0001, 0.0001< *** p < 0.001, 0.001 < **p < 0.01, 0.01 < * p < 0.05)

Analysis of neurotrophic factors and neurotransmitters in the blood before and after the intervention

Neurotrophic factor and neurotransmitter results in the blood before and after the intervention

After the intervention, the concentrations of BDNF, NGF, DA, GABA, 5-HT, and ACh in the blood of the three groups of patients increased significantly (p < 0.05), but the concentration of Glu in the combined group decreased significantly (p < 0.05). The time × group interaction was only significant (p < 0.05) for the concentrations of BDNF, NGF, 5-HT, and Glu, and after the intervention, the concentration of BDNF in the combined group was significantly greater than that in the tDCS group (p = 0.005) and the AE group (p < 0.001); the concentration of NGF was significantly greater than that in the tDCS group (p < 0.001) and the AE group (p = 0.033). These findings reveal that the improvement in BDNF and NGF concentrations in the combined group was more significant than those in the tDCS group and the AE group, as shown in Table 7; Fig. 10.

Table 7.

Analysis of the levels of neurotrophic factors and neurotransmitters in the blood of the three groups of patients (n = 74)

Neurotrophic factors Neurotransmitters	Group	Before the intervention	After the intervention	F1,71	p	η2p
BDNF (ng/mL)	tDCS + AE group	2.66 ± 1.11	4.21 ± 0.98	57.229	<0.001	0.446
	tDCS group	2.69 ± 0.96	3.30 ± 1.02	8.808	0.004	0.110
	AE group	2.06 ± 0.71	3.01 ± 0.92	20.581	<0.001	0.225
	F 2,71	3.452	10.117
	p	0.037	<0.001
	η 2 p	0.089	0.222
Time (F2,71, p, η2p)	75.588	<0.001	0.516
Group (F2,71,p,η2p)	7.492	0.001	0.174
Time×Group (F2,71, p, η2p)	5.415	0.006	0.132
NGF (pg/mL)	tDCS + AE group	37.84 ± 17.16	68.92 ± 10.78	72.074	<0.001	0.504
	tDCS group	37.23 ± 17.02	52.23 ± 13.27	16.770	<0.001	0.191
	AE group	38.67 ± 9.80	59.98 ± 11.73	32.537	<0.001	0.314
	F 2,71	0.056	12.165
	p	0.946	<0.001
	η 2 p	0.002	0.255
Time (F2,71, p, η2p)	111.387	<0.001	0.610
Group (F2,71,p,η2p)	4.582	0.013	0.114
Time×Group (F2,71, p, η2p)	4.900	0.010	0.121
DA (pg/mL)	tDCS + AE group	294.90 ± 83.05	420.66 ± 129.86	16.897	<0.001	0.192
	tDCS group	267.83 ± 73.40	333.18 ± 110.62	4.563	0.036	0.060
	AE group	316.89 ± 108.57	384.37 ± 139.75	4.670	0.034	0.062
	F 2,71	1.859	2.986
	p	0.163	0.057
	η 2 p	0.050	0.078
Time (F2,71, p, η2p)	23.489	<0.001	0.249
Group (F2,71,p,η2p)	3.899	0.025	0.099
Time×Group (F2,71, p, η2p)	1.248	0.293	0.034
GABA (pg/mL)	tDCS + AE group	131.75 ± 89.45	176.49 ± 78.28	8.352	0.005	0.105
	tDCS group	100.66 ± 45.27	136.41 ± 40.53	5.334	0.024	0.070
	AE group	164.57 ± 97.31	162.37 ± 37.39	0.019	0.890	0.000
	F 2,71	3.868	3.352
	p	0.025	0.041
	η 2 p	0.098	0.086
Time (F2,71, p, η2p)	8.410	0.005	0.106
Group (F2,71,p,η2p)	4.241	0.018	0.107
Time×Group (F2,71, p, η2p)	2.522	0.087	0.066
Glu (ng/mL)	tDCS + AE group	2921.89 ± 1049.10	2219.01 ± 658.53	7.699	0.007	0.098
	tDCS group	2454.35 ± 896.24	2686.91 ± 1047.10	0.843	0.362	0.012
	AE group	2918.62 ± 1906.9 6	3184.50 ± 1604.65	1.058	0.307	0.015
	F 2,71	0.987	4.223
	p	0.378	0.018
	η 2 p	0.027	0.106
Time (F2,71, p, η2p)	0.214	0.645	0.003
Group (F2,71,p,η2p)	1.583	0.213	0.043
Time×Group (F2,71, p, η2p)	4.678	0.012	0.116
5-HT (ng/mL)	tDCS + AE group	224.61 ± 131.06	369.98 ± 141.72	100.653	<0.001	0.586
	tDCS group	215.13 ± 150.15	284.58 ± 138.89	22.974	<0.001	0.244
	AE group	208.34 ± 126.49	300.27 ± 147.73	38.641	<0.001	0.352
	F 2,71	0.088	2.527
	p	0.916	0.087
	η 2 p	0.002	0.066
Time (F2,71, p, η2p)	147.344	<0.001	0.675
Group (F2,71,p,η2p)	0.939	0.396	0.026
Time×Group (F2,71, p, η2p)	7.233	0.001	0.169
Ach (pg/mL)	tDCS + AE group	72.26 ± 25.94	127.77 ± 38.58	54.020	<0.001	0.432
	tDCS group	62.03 ± 25.50	117.03 ± 29.79	53.033	<0.001	0.428
	AE group	87.01 ± 40.29	124.90 ± 48.55	24.164	<0.001	0.254
	F 2,71	3.961	0.493
	p	0.023	0.613
	η2p	0.100	0.014
Time (F2,71, p, η2p)	126.934	<0.001	0.641
Group (F2,71,p,η2p)	1.863	0.163	0.050
Time×Group (F2,71, p, η2p)	1.715	0.187	0.046

Fig. 10.

Time (before the intervention vs. after the intervention) × group (tDCS + AE vs. tDCS vs. AE) interaction effects of the blood neurotrophic factor and neurotransmitter levels in the three groups before and after the intervention. A: BDNF; B: NGF; C: DA; D: GABA; E: Glu; F: 5-HT; G: ACh (****p < 0.0001, 0.0001< *** p < 0.001, 0.001 < **p < 0.01, 0.01 < * p < 0.05)

Correlations between the changes in neurotransmitter levels in the blood and behavioral changes

After performing Spearman correlation analysis on the changes in neurotransmitter levels in the blood before and after the intervention and the behavioral changes, the results showed that after Bonferroni correction, the changes in BDNF concentrations in the blood were negatively correlated with the differences in Stroop congruent reaction times (r = -0.405, p < 0.00034), the changes in NGF concentrations were positively correlated with the differences in MMSE scores (r = -0.405, p < 0.00034), whereas the other indicators were not correlated. These findings indicate that an increase in BDNF concentration is associated with an increase in congruent reaction time in the inhibitory control task in terms of executive function and that an increase in NGF concentration is associated with an improvement in cognitive function, as shown in Figs. 11 and 12.

Fig. 11.

Heatmap of the correlation analysis results of the changes in nerve growth factor and neurotransmitter levels in the blood and the behavioral changes

Fig. 12.

Scatter plot of the correlation analysis results of the changes in nerve growth factor and neurotransmitter levels in the blood and the behavioral changes

A: Scatter plot of the correlation between the changes in BDNF concentrations and the differences in Stroop congruent reaction times; B: Scatter plot of the correlation between the changes in NGF concentrations and the differences in MMSE scores.

Discussion

This study used 2.0 mA anodal tDCS of the left DLPFC combined with moderate-intensity AE as an intervention in the enrolled PSCI patients. After clinical scales and executive function task paradigms were evaluated, compared with tDCS stimulation or the AE intervention alone, the combined intervention more effectively improved the executive function of PSCI patients to a certain extent. Further testing of the concentrations of neurotrophic factors and neurotransmitters in the blood revealed that the combination of the two significantly changed the concentrations of some related neurotrophic factors in the blood. Notably, these observations were based on sampling conducted one day after the intervention, indicating that the detected changes reflect sustained adaptive effects rather than acute transient responses.

Effects of the combined tDCS and AE intervention on executive and cognitive function in patients with PSCI

This study revealed that compared with either tDCS or AE alone, the combined tDCS and AE intervention resulted in greater improvement in executive function (as measured by the C-EXIT25 scores) in patients with PSCI. These findings align with the existing evidence from Pahlman et al. [33], who followed 74 stroke patients and reported that patients with executive dysfunction had a fourfold greater risk of poor motor recovery one year poststroke, underscoring that targeted improvement in executive function may optimize patient prognosis. The enhanced efficacy of the combined intervention may stem from the potential synergistic mechanisms of neuromodulation between the two approaches. At the neurophysiological level, tDCS and AE may influence executive function networks through distinct yet complementary pathways. Prior studies have shown that anodal tDCS over the left DLPFC can enhance functional connectivity within the default mode and salience networks, thereby improving cognitive capacity [34, 35]. Concurrently, acute AE has been shown to immediately activate the left DLPFC and enhance integration between attention and executive control networks [36, 37]. Therefore, the combined intervention may produce additive or synergistic effects by simultaneously modulating cortical excitability and network coherence, leading to more effective improvement in executive function. However, compared with single-modality interventions, the combined intervention did not demonstrate superior effects on global cognitive function according to the MMSE scores. These findings suggest that the cognitive benefits of combined tDCS and AE may be domain specific and particularly focused on executive function. Future studies should consider the impact of stroke lesion location on cognitive outcomes and perform stratified analyses based on stroke location or type for different cognitive domains [38].

Further analysis using specific task-based assessments revealed that the effects of the combined intervention were selective across executive subdomains. In the inhibitory control task, the combined intervention significantly improved reaction times in the congruent condition of the Stroop task, but no significant improvement was observed in the incongruent condition or interference effect. The left DLPFC is a key region involved in the Stroop effect, and tDCS over this area has been shown to enhance performance in healthy individuals [39]. AE has also been shown to alter activation patterns in the anterior cingulate cortex during Stroop task performance [40]. The combined approach may integrate these benefits, comodulating relevant brain regions to improve processing efficiency in the congruent condition, which relies more on automated processing and basic response speed. However, for the incongruent condition, which requires stronger inhibitory control, the current combined protocol may have limited effects. In the working memory task, the combined intervention led to greater improvement in verbal working memory in PSCI patients, specifically increasing the hit accuracy in the 1-back task. However, no significant improvement was observed in the correct rejection accuracy or delayed word recall. The DLPFC plays a central role in working memory [41], and both tDCS [42] and AE [43] have been shown to enhance working memory performance individually. The current findings suggest that the combined intervention may be more effective at enhancing short-term memory maintenance and simple updating, whereas processes requiring greater amounts of cognitive resources, such as interference suppression (correct rejection) and long-term retrieval (delayed recall), showed limited improvement. This may be related to the characteristic executive dysfunction in PSCI, where patients have difficulty efficiently processing and retaining short-term information [44]. The combined intervention only partially improved performance on the cognitive flexibility task TMT-A (primarily reflecting visual search and motor speed), with no significant effect on TMT-B, which better assesses cognitive flexibility. This finding indicates that compared with higher-order cognitive flexibility, the current combined protocol may be more effective at enhancing basic processing speed.

This study also found that the combined intervention resulted in a greater reduction in HAMD scores among PSCI patients. tDCS has been established as an effective intervention for poststroke depression [45] and treatment-resistant depression [46]. Although AE alone did not independently improve depression scores in this study, structured aerobic training has been shown to alleviate depressive symptoms and improve quality of life in stroke patients over time [47]. Furthermore, studies on conditions such as fibromyalgia have directly demonstrated that combining tDCS (targeting the left M1 region) with AE can effectively improve depressive mood. This finding suggests that the combined intervention may have synergistic effects on depressive symptoms in PSCI patients through multiple pathways involving neuromodulation and physiological/psychological regulation.

Additionally, no significant improvements in motor function or activities of daily living were observed following the combined intervention. Although improvements in executive function and motor recovery may be interrelated [48], the relatively short observation period and the use of a single-mode AE intervention (lower limb cycling) in this study may have been insufficient to induce broad functional changes. Future studies should consider incorporating varied AE modalities and intensities, along with longer follow-up periods, to comprehensively evaluate the effects of combined tDCS and AE on multidimensional functional outcomes in PSCI patients.

Effects of tDCS combined with AE on neurotrophic factors and neurotransmitters in the blood of PSCI patients

BDNF plays a pivotal role in the onset and progression of PSCI. Studies have demonstrated that, compared to stroke patients without cognitive impairment, the blood concentration of BDNF in PSCI patients is significantly reduced, and this reduction is strongly correlated with cognitive impairment [49]. In this study, it was found that all three intervention methods (tDCS alone, AE alone, and combined tDCS and AE) effectively increased the blood concentration of BDNF. The combined intervention group showed a more significant improvement compared to the single tDCS group and the AE group. Moreover, the increase in BDNF was associated with a shortening of the Stroop congruent reaction time. BDNF is an important neurotrophic factor that is widely distributed in the central nervous system, particularly in brain regions closely related to learning and memory, such as the hippocampus and prefrontal cortex [21]. Additionally, BDNF is also expressed in skeletal muscle and, as a member of the myogenic neurotrophic factors, plays a regulatory role in the neuromuscular system [50]. Studies have shown that a 2.0 mA, 20-minute anodal tDCS intervention applied 14 times can improve the MMSE scores of PSCI patients and significantly increase the blood concentration of BDNF [51]. Fritsch et al. [52] found that individuals with the BDNF Val66Met polymorphism, which affects BDNF secretion, exhibit lower motor skill acquisition and executive function improvement after tDCS intervention, indicating that BDNF secretion is crucial for tDCS to enhance executive function. At the same time, BDNF is an exercise-dependent neurotrophic factor, and exercise is an effective means to increase BDNF concentration, improve neural circuits, and enhance cognitive function [53]. Studies have found that AE is the main source of BDNF-dependent enhancement of executive function [54]. Exercise-induced upregulation of BDNF can protect neurons, increase neuroplasticity, and promote neuronal survival and functional recovery. Hillman et al. [55] and Leckie et al. [56], in their studies on subjects of different ages, found that exercise intervention led to changes in serum BDNF expression. This change can, to a certain extent, mediate the relevant brain pathways required for exercise and specific executive function performance, as well as “top-down” cognitive processing. Electrophysiological assessments of brain function also support this argument. Meanwhile, Pan et al. [57] demonstrated that BDNF can undergo bidirectional transport across the blood‒brain barrier; Karege et al. [58] further revealed that serum BDNF levels are positively correlated with BDNF concentrations in the cortex and hippocampus. Wrann et al. [53] elucidated the “muscle-brain axis” signaling pathway, whereas Rasmussen et al. [59] provided evidence that the brain itself releases BDNF into the bloodstream during exercise. The combined intervention of tDCS and AE may synergistically promote BDNF expression through distinct mechanisms. Specifically, tDCS directly modulates central neural activity via electrical stimulation, whereas AE indirectly enhances neuronal function through physiological activity. The integration of these two interventions is expected to further upregulate BDNF levels, thereby significantly improving executive function. This central BDNF upregulation, facilitated by the bidirectional permeability of the blood‒brain barrier and amplified by exercise-induced peripheral-central crosstalk, provides a solid neurobiological basis for the observed synergistic effect. The combined intervention may comprehensively regulate neuronal growth, differentiation, and survival, enhance synaptic plasticity, and promote neurotransmitter release, thereby achieving improvements in executive function. Notably, a previous study in healthy young individuals also found that tDCS combined with AE intervention significantly increased peripheral blood BDNF levels [24]. Based on similar phenomena observed in patients with PSCI, it is plausible that this mechanism represents the potential pathway through which the combined intervention improves executive function in PSCI patients, a conclusion that is highly consistent with the review by Steinberg et al. [21].

NGF was the first neurotrophic factor to be discovered and plays important roles in the growth, survival, and plasticity of neurons. Studies have shown that changes in NGF levels after stroke may affect neural repair and cognitive function recovery [60]. Like BDNF, NGF is crucial for the growth, survival, and plasticity of neurons [61]. In this study, compared with the tDCS group or the AE group, all three interventions increased the blood concentration of NGF, but the combined intervention group showed a more significant increase. Moreover, the change in NGF concentration was positively correlated with the change in MMSE score. Studies have shown that after tDCS, the concentrations of BDNF and NGF in the blood of PSCI patients increase significantly, accompanied by improved cognitive function [62]. Similarly, AE can not only increase the expression of NGF in humans [63], but studies in type 1 diabetic rat models have also shown that 8 weeks of treadmill exercise can improve the learning and memory ability of type 1 diabetic rats by increasing the expression of hippocampal NGF and improving anti-inflammatory ability [64]. These findings indicate that when tDCS and AE are used in combination, more significant therapeutic effects may be achieved through the synergistic effects of multiple signaling pathways. tDCS increases the excitability of cortical neurons and the expression of NGF by directly stimulating the cerebral cortex, whereas AE promotes the expression of NGF and the functional recovery of neural networks by increasing cerebral blood flow and oxygen supply, which regulates the growth, differentiation, and survival of neurons, promotes synaptic plasticity and neurotransmitter release, and subsequently improves executive and cognitive functions. The combination of tDCS and AE may maximize the role of promoting an increase in NGF, producing a stronger synergistic effect and improving the overall effect of the intervention.

This study revealed a significant decrease in blood Glu levels after the combined tDCS and AE intervention (no intergroup significance), indicating that the combined intervention has regulatory effects on Glu levels. Supporting evidence includes the following: tDCS selectively modulates GABAergic/glutamatergic neurotransmission to protect neurons and improve neural function [65]; exercise acutely regulates cortical Glu/GABA in healthy adults, enhancing visual cortex Glu/GABA levels and attention [66]. We hypothesize that the combined intervention induces complex, multilevel neurotransmitter regulation to reduce Glu levels—which is beneficial for PSCI through the optimization of neural adaptive regulation and the mitigation of oxidative stress and synaptic excitotoxicity. Consistent with these findings, 8 weeks of enriched environment rehabilitation improved PSCI patients’ cognitive/executive function and reduced blood Glu levels, potentially by enhancing synaptic plasticity and alleviating oxidative stress/inflammation [67].

Numerous studies have confirmed that 5-HT, DA, GABA and ACh contribute to improvements in cognition and emotions in patients with PSCI [68–70]. In this study, all three intervention groups presented elevated 5-HT, DA, GABA and ACh levels, but the combined group did not exhibit superior efficacy over the single-intervention groups. This may result from tDCS or AE alone already maximizing the changes in the concentrations of these neurotransmitters, precluding further secretion enhancement by combination.

Limitations

In future studies of the effects of tDCS combined with AE on executive function in PSCI patients, the sample size should be increased, and multicenter randomized controlled trials should be conducted to more comprehensively verify the effectiveness and safety of the combined intervention in PSCI patients in terms of executive function and broader cognitive areas. Moreover, given that this study only used a combination of 2.0 mA intensity anodal tDCS of the left DLPFC and 30 min of moderate-intensity aerobic cycling for 4 weeks, subsequent studies should further explore the effects of different tDCS and AE treatment sequences, stimulation parameters (such as intensity and duration), exercise programs (such as intensity, duration, and type) and other factors on the intervention effect and conduct long-term follow-up evaluations to further explore the optimal combination of tDCS and AE.

Future directions

On the basis of the results of the current study, this study provides a foundation for further exploration of the role of central control technology combined with peripheral control technology in executive function and cognitive dysfunction-related diseases. Future research can focus on the synergy of more central control technologies, such as tDCS and TMS, and peripheral control technologies, such as different AE modes, to further explore their potential in the treatment of related diseases. In subsequent studies, clinical trial plans can be further optimized, the relevant parameters of various central control technologies can be clarified, and a variety of AE modes and intensities can be combined to form a diversified treatment combination to comprehensively evaluate its improvement effect on executive function and cognitive function.

Author contributions

Kai Zheng, Ying Shen, Haohao Zhu and Yingying Ji conceived the study; Yingying Ji, Yan Han, Huimin Zhou and Qian Lu performed the survey and summary; and Yingying Ji, Yan Han, Haohao Zhu and Huimin Zhou wrote and revised the manuscript. All the authors have approved the publication of the manuscript.

Funding

This work is supported by the National Natural Science Foundation of China (82571738), the Basic Research Program of Jiangsu (BK20250305, BK20251755), the Key Project of Jiangsu Province's Key Research and Development Program (BE2023023-2), the Health International (Regional) Exchange Support Program of Jiangsu Province, the Jiangsu Provincial Health Commission General Project (H2023040), the Top Talent Support Program for Young and Middle-aged People of Wuxi Health Committee (HB2023086), the Wuxi Municipal Health Commission (Q202167), and the Wuxi Taihu Talent Project (WXTTP2021).

Data availability

The dataset generated and analyzed during the current study is available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Wuxi Mental Health Centre, under grant number WXMHCIRB2021LLKY145.

Consent for publication

The patients were informed of the study’s purpose and signed an informed consent form.

Competing interests

The authors declare no competing interests.