Combined home-delivered transcranial direct current stimulation and cognitive training in older adults with mild cognitive impairment due to Alzheimer’s disease: a randomized, single-blind, sham-controlled trial

Abstract

Background

Non-pharmacological interventions such as transcranial direct current stimulation (tDCS) and cognitive training (CT) may benefit cognition in older adults with mild cognitive impairment due to Alzheimer’s disease (MCI-AD), yet evidence on combined, home-delivered protocols remains limited. This study evaluated the effectiveness of a combined home-delivered combined tDCS and CT intervention across multiple cognitive domains in individuals with MCI-AD.

Methods

We conducted a randomized, single-blind, sham-controlled trial in 41 participants (≥ 65 years) with MCI-AD, allocated to four experimental conditions: (1) tDCS combined with CT, (2) tDCS without CT, (3) sham tDCS combined with CT, and (4) sham tDCS without CT. Interventions were delivered at participants’ homes by trained staff across 10 daily sessions. Outcomes (MMSE; TAVEC Trial 1/Total/Delayed; Forward Digit Span) were assessed at baseline, post-intervention, and 1-month follow-up. Data were analyzed with 4 × 3 mixed-design ANOVAs (group×time) with Bonferroni-adjusted post hoc tests; partial eta-squared was reported as effect size.

Results

Significant group × time interactions were found for global cognition (MMSE), immediate memory (TAVEC Trial 1), verbal learning (TAVEC Total), and short-term memory (Forward Digit Span) (all p<.05). The combined tDCS + CT group showed sustained improvements across the significant cognitive domains at post-intervention and follow-up.

Conclusions

This randomized controlled trial provides evidence that a home-delivered combined intervention of tDCS and CT can enhance cognitive performance in older adults with MCI-AD. Improvements were domain-specific and sustained over short follow-up. While the modest sample and limited follow-up warrant cautious interpretation, the findings support the potential of combined non-pharmacological interventions delivered at home and highlight the need for larger longitudinal studies to establish their long-term efficacy and scalability.

Trial registration

ClinicalTrials.gov, NCT06861231. Registered on February 28, 2025.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-026-07481-z.

Background

Mild cognitive impairment due to Alzheimer’s disease (MCI-AD) represents an early clinical stage of Alzheimer’s pathology, formally defined by the National Institute on Aging-Alzheimer’s Association (NIA-AA) as a condition characterized by cognitive decline greater than expected for age and education, preserved functional independence, and high likelihood of Alzheimer’s etiology [1]. Epidemiological data from the Alzheimer’s Association indicate that approximately 8–11% of adults aged 65 and older present MCI-AD, equivalent to about 5–7 million individuals in the United States [2]. MCI-AD follows an insidious and progressive course. In clinical samples, annual conversion rates to dementia have been estimated at around 13%, compared to ~ 3% in community-based cohorts [3]. More recent meta-analytic evidence reports typical annual conversion rates of 10–15% [4], in contrast to much lower rates in cognitively healthy individuals. Positioned between normal aging and dementia, this stage is increasingly recognized as a strategic window for early intervention aimed at slowing down the trajectory of cognitive decline and preserving functional abilities [5].

Despite ongoing pharmacological advancements, available treatments for Alzheimer’s Disease remain limited in their ability to modify the early course of the disease and provide only modest symptomatic benefits, even in individuals with MCI-AD [6]. As a result, non-pharmacological therapies have gained increasing recognition as complementary, evidence-based strategies to support cognitive and functional outcomes in the early phases of the disease [7, 8]. Among these interventions, cognitive training is one of the most extensively studied and promising approaches for enhancing cognitive function in individuals with MCI-AD [9].

Cognitive training (CT) is typically implemented through structured and repetitive cognitive practice, delivered in-person or at home, and can involve trained caregivers to support adherence and effectiveness [10]. Its primary goal is to enhance or maintain cognitive functions and, indirectly, to support everyday functioning. At the neurocognitive level, CT is hypothesized to stimulate cognitive reserve and promote neuroplasticity in brain networks essential for memory and executive function, particularly the frontoparietal network and hippocampus [11].

Evidence from randomized controlled trials, systematic reviews, and recent meta-analyses indicates that CT produces modest but reliable benefits across multiple cognitive domains in individuals with MCI and early Alzheimer’s disease, including global cognition, episodic memory, attention, and learning, with effects that may persist beyond six months and, in some cases, up to five years [12–15]. Although most studies included MCI of mixed etiology, these findings are highly relevant to MCI-AD, given the predominance of Alzheimer’s pathology in amnestic MCI. Moreover, available evidence suggests that individual factors such as baseline cognitive status and age may modulate responsiveness to CT [16]. In patients with early-stage Alzheimer’s disease, clinical trials indicate that CT can stabilize global cognitive performance and improve delayed memory, compared with untreated controls [12], with emerging neuroimaging evidence supporting training-induced functional connectivity changes consistent with neuroplastic mechanisms [17–18].

Given the growing interest in enhancing the efficacy of cognitive interventions, non-invasive brain stimulation techniques such as transcranial direct current stimulation (tDCS) have gained attention as potential modulators of neuroplasticity and cognitive performance. tDCS delivers low-intensity electrical currents to the scalp and modulates cortical excitability in a polarity-dependent manner, thereby influencing the likelihood of neuronal activation without directly inducing action potentials [19]. At the neurobiological level, tDCS has been associated with mechanisms related to synaptic plasticity, including facilitation of long-term potentiation [20, 21].

Clinically, evidence from randomized trials and meta-analyses indicates that tDCS can produce modest but meaningful improvements in cognitive performance in older adults and individuals with MCI or early Alzheimer’s disease, particularly when stimulation is applied over prefrontal regions such as the dorsolateral prefrontal cortex (DLPFC) [22–25]. Importantly, the magnitude of these effects appears greater in earlier disease stages, where neuroplastic mechanisms are relatively preserved [24].

The combination of CT with tDCS is grounded in the principle of activity-dependent plasticity, which posits that externally applied stimulation is more effective when paired with active cognitive engagement. Rather than producing merely additive effects, CT and tDCS are thought to interact synergistically, with stimulation enhancing learning-related modulation within neural circuits engaged by the training tasks [26].

When applied over prefrontal regions, tDCS can increase cortical excitability and facilitate synaptic plasticity, creating a neurophysiological context that supports the effects of structured CT [21, 27]. In this context, tDCS acts as a facilitator rather than a direct cognitive enhancer, amplifying task-related neural activation and promoting plastic changes within relevant networks. This interaction provides a strong rationale for combining both approaches in interventions targeting populations at risk for cognitive decline [26, 28].

Empirical evidence supports the efficacy of combining CT with tDCS to enhance cognitive performance in older adults, particularly in domains of global cognition and memory. In a randomized controlled trial, the authors [29] found that older adults receiving combined CT and anodal tDCS over the left dorsolateral prefrontal cortex exhibited greater improvements in memory performance and cortical network connectivity compared to those receiving CT alone. Consistently, a meta-analysis reported small-to-moderate effect sizes for improvements in global cognition and episodic memory following combined interventions, with the most consistent effects observed in individuals with MCI [27].

It has been demonstrated that the pairing of tDCS with active cognitive engagement enhances episodic memory more reliably than stimulation alone, and may support the maintenance of training-related gains beyond the intervention period [25]. This suggests that tDCS may facilitate both immediate and sustained cognitive benefits, potentially through the reinforcement of synaptic mechanisms supporting memory retention. Although outcomes vary depending on individual characteristics and intervention protocols, current evidence indicates that combined approaches tend to produce more consistent and durable cognitive benefits than either intervention in isolation, particularly in populations with preserved neuroplastic potential. These findings are especially relevant for early Alzheimer’s disease, where delaying cognitive decline may help preserve autonomy and reduce long-term care demands [30].

In light of the empirical findings outlined above, the present study targets individuals with MCI-AD, an early stage of the disease in which cognitive deficits emerge but everyday functional abilities remain relatively intact. Given the growing interest in early non-pharmacological interventions, identifying protocols that yield stronger and more durable cognitive benefits may have important clinical implications for slowing functional decline. Beyond cognitive outcomes, such approaches may also offer practical advantages in terms of cost-effectiveness and accessibility. Non-pharmacological interventions like CT and tDCS can be delivered in outpatient or home settings with minimal side effects, making them scalable and adaptable to different clinical environments.

Accordingly, the aim of this study was to compare four experimental conditions derived from the combination of transcranial direct current stimulation (tDCS; real vs. sham) and cognitive training (CT; present vs. absent), in order to examine their individual and interactive effects on global cognition and memory performance in older adults with MCI-AD. Based on models of activity-dependent plasticity and prior evidence suggesting synergistic effects between CT and non-invasive brain stimulation, we hypothesized that the combined tDCS + CT condition would produce greater improvements in global cognition and memory outcomes than either intervention alone or sham stimulation conditions.

Materials and methods

Study design and participants

We conducted a randomized, single-blind, sham-controlled, parallel-group, superiority trial with 1:1:1:1 allocation involving a 10-day home-delivered intervention. Participants with MCI-AD, diagnosed according to the National Institute on Aging - Alzheimer’s Association (NIA-AA) criteria [1], were assessed at baseline, immediately after the intervention, and at a 1-month follow-up. Participants were randomly assigned to 1 of 4 conditions: tDCS with the anode positioned over the left dorsolateral prefrontal cortex (DLPFC), with or without CT, or sham tDCS with or without CT. Eligibility was confirmed during the baseline assessment prior to randomization.

Participants were required to meet the following inclusion criteria: (1) age over 65 years; (2) a score of 3 on the Global Deterioration Scale (GDS) [31] consistent with mild cognitive impairment (MCI); and (3) fulfilment of clinical and cognitive criteria for MCI-AD, as defined by the NIA-AA guidelines, including: (a) evidence of progressive cognitive decline, particularly in memory and learning; (b) relatively preserved independence in activities of daily living [32]; and (c) no other medical or psychiatric conditions explaining the symptoms. The age threshold of 65 years was selected to focus on Alzheimer’s disease in older adults and to reduce clinical heterogeneity by excluding cases of early-onset cognitive impairment, which may differ in etiology and clinical course.

As part of the screening process, the Memory Alteration Test (M@T) [33] and the Mini-Mental State Examination (MMSE) [34] Spanish validation by Lobo et al. [35], were administered. The M@T was used to support cognitive classification and apply exclusion thresholds: individuals with scores ≤ 27 (suggestive of dementia) or ≥ 37 (indicative of normal cognition) were excluded, prioritizing those with intermediate scores consistent with MCI. The MMSE was first used to exclude individuals with moderate to severe cognitive impairment (cut-off < 20). It was subsequently employed as the primary outcome measure of global cognitive function. To mitigate potential circularity, that is, the risk of overestimating intervention effects or reducing the validity of outcome measurements due to overlap between selection criteria and outcome assessment, the MMSE was not used to stratify or select participants within the eligible range, and its use as an outcome focused on change over time rather than absolute scores. This approach was justified by the well-established clinical relevance and sensitivity of the MMSE to detect cognitive change over time in MCI populations [36].

Additional exclusion criteria included diagnosis of neurodegenerative diseases other than Alzheimer’s disease, severe psychiatric symptoms, significant functional dependency, contraindications for tDCS (e.g., intracranial metallic implants or intracranial hypertension), or plans to relocate during the intervention period.

Participants were recruited from the Neurology Department of the Consorcio Hospital General Universitario of Valencia (Spain). Participants had undergone prior clinical diagnostic evaluation by neurologists at the recruiting hospital. The diagnosis of MCI due to Alzheimer’s disease was established before study inclusion, and the assessments conducted in the present study were used solely to confirm eligibility and disease stage, not to determine etiology. Initially, 65 patients were contacted. Of these, 17 declined to participate, 4 were excluded for scoring below 20 on the MMSE, and 3 withdrew before completing the intervention period. The final sample consisted of 41 participants with MCI-AD who met eligibility criteria.

The study was approved by the Human Research Ethics Committee of the University of Valencia (Spain). All participants received a complete description of the procedure and then signed written informed consent prior to inclusion. A flowchart of participant progression is shown in Fig. 1.

Fig. 1.

Flowchart of participant recruitment, allocation, and retention

Following baseline assessment and confirmation of eligibility, participants were randomly assigned to one of four experimental groups: (1) tDCS without CT (n = 11), (2) tDCS with CT (n = 11), (3) sham tDCS with CT (n = 9), and (4) sham tDCS without CT (n = 10). Stratified block randomization by gender was applied to ensure balanced allocation across groups and minimize baseline differences. The randomization sequence was computer-generated using permuted blocks (block size = 4) with a 1:1:1:1 allocation ratio. Allocation was concealed using sequentially numbered, opaque, sealed envelopes. Investigators enrolled participants; group assignment was performed by a trial coordinator with no access to outcome data. Participants were blinded to group allocation throughout the intervention, consistent with the single-blind design. Outcome assessors were blinded to group allocation throughout the study.

The overall sample had a mean age of 75.54 years (SD = 5.41), ranging from 66 to 86 years, and was gender-balanced (21 men and 20 women). The average level of education was 10.49 years (SD = 3.79; range: 5–20 years).

Table 1 summarizes the demographic and baseline cognitive characteristics of the 4 groups. No statistically significant differences were found between groups in age, gender, years of education, or baseline cognitive performance, suggesting comparable conditions at study entry.

Table 1.

Baseline demographic and cognitive characteristics of the experimental groups

	tDCS (n = 11)	tDCS + CT (n = 11)	Sham (n = 10)	Sham + CT (n = 9)	p
Age (M ± SD)	75.09 ± 4.09	76.18 ± 6.10	74.10 ± 5.07	76.89 ± 6.68	0.698
Gender (M/W)	6/5	6/5	5/5	4/5	0.966
Education (years)	10.00 ± 3.66	10.55 ± 2.88	9.80 ± 4.05	11.78 ± 4.84	0.685
MMSE (M ± SD)	25.09 ± 1.58	25.18 ± 1.66	24.70 ± 2.06	24.89 ± 1.76	0.927
M@T (M ± SD)	32.27 ± 2.83	32.36 ± 2.20	33.10 ± 2.89	32.78 ± 3.11	0.895

Measures

Eligibility measures

Participant eligibility was assessed using the Mini-Mental State Examination (MMSE) [34], Spanish validation by Lobo et al. [35], the Memory Alteration Test (M@T) [33] the Global Deterioration Scale (GDS) [31], and the Lawton and Brody Instrumental Activities of Daily Living Scale [32].

The MMSE is a widely used screening tool that evaluates global cognitive functioning, with a maximum score of 30 points. In this study, a score below 21 was considered indicative of moderate to severe cognitive impairment and led to exclusion. The M@T is a brief neuropsychological test that assesses multiple memory-related domains, including encoding, orientation, semantic memory, and free recall. It was used during screening to support cognitive classification and apply exclusion thresholds based on performance range. GDS is a clinician-rated scale designed to classify the overall stage of cognitive decline, ranging from normal aging (Stage 1) to severe dementia (Stage 7). In this study, the GDS was used to support the clinical classification of MCI. Only participants scoring Stage 3 were included, a stage characterized by objective cognitive decline, typically affecting memory, without significant impairment in daily functioning. To assess functional status, Instrumental Activities of Daily Living Scale was also administered. This tool evaluates the capacity to perform eight everyday activities required for independent living. Scores range from 0 (fully dependent) to 8 (fully independent), and individuals with high levels of dependency were excluded from participation.

Primary outcome measures

The selection of primary cognitive outcome domains was guided by their clinical relevance in MCI-AD. Global cognitive functioning was included to capture overall cognitive change associated with disease progression, while verbal episodic memory and learning were selected given their early impairment and central role in diagnostic characterization. Immediate verbal memory span was included as an index of short-term and working memory processes, which are frequently targeted in CT protocols and have shown responsiveness to prefrontal tDCS in prior studies. Together, these domains represent core cognitive functions affected in early Alzheimer’s disease.

Primary outcomes included global cognitive functioning, verbal episodic memory (immediate, learning, and delayed), and immediate verbal memory span. These measures were collected at baseline, post-intervention, and 1-month follow-up.

The MMSE, beyond its role in initial screening, served as the primary outcome measure of global cognitive functioning. This Spanish validation is widely used in clinical and research contexts in Spain. MMSE was used as an index of global cognitive functioning; the maximum score is 30 points. This test was designed to estimate the existence and severity of cognitive impairment; it is a brief and quantitative test that measures general cognitive function.

Learning Test (TAVEC) [37]. This test presents a list of 16 words that, after being read by the evaluator, have to be repeated by the participant. The list is repeated 5 times (trials), and after 20 min, the participant is again asked to remember the 16 words. Immediate memory was measured using the first trial, learning ability with the fifth trial and the sum of the five trials, and delayed memory through free recall after a 20-minute interval.

Finally, immediate verbal memory span was measured using the forward digit span subtest of the Wechsler Adult Intelligence Scale-III (WAIS-III) [38]. In this task, participants are asked to repeat sequences of digits in the same order presented by the examiner. One point is awarded for each correct sequence, with a maximum score of 16.

Interventions

All interventions were conducted at participants’ homes by trained members of the research team, who visited daily over a 10-day period to administer the assigned condition (tDCS or sham tDCS, with or without CT). This ensured strict adherence to the protocol and correct application of stimulation and training procedures. Depending on group allocation, participants received either tDCS or sham tDCS, with or without CT. In the combined conditions, tDCS was applied online. CT started simultaneously with tDCS and continued throughout the 20-minute tDCS period. After the end of tDCS, participants continued with the CT tasks until completing the full training session.

Transcranial direct current stimulation (tDCS)

tDCS was delivered using a battery-operated HDCStim device (Newronika™, Milan, Italy), which provides non-invasive direct current stimulation. Each session lasted 20 min, with a stimulation intensity of 2 mA and a current density of 0.08 mA/cm². Electrodes consisted of 5 × 5 cm saline-soaked sponges, designed to optimize conductivity and reduce skin resistance. Electrodes were positioned on a neoprene cap following the 10–10 international EEG system: the anode was placed at F3 (left dorsolateral prefrontal cortex, DLPFC), and the cathode at Fp2 (right supraorbital area). This electrode montage continues to be commonly used in studies of cognitive enhancement, with a recent meta-analysis highlighting the influence of optimal dosing parameters [24]. Electrode-skin impedance was checked prior to each session to ensure adequate current delivery. Stimulation was initiated only when impedance values were within the acceptable range.

Each stimulation session included a 30-second ramp-up and ramp-down phase to facilitate sensory habituation. The sham condition replicated all technical and procedural aspects of real tDCS delivery, including ramping, but the current was only applied during the initial and final 30 s. This approach ensured participants experienced the characteristic tingling sensation, thereby supporting effective blinding. The sham protocol was therefore indistinguishable from real tDCS in terms of procedure and device appearance, and participants were unable to determine their group assignment.

Cognitive training (CT)

The CT program was based on a protocol developed for individuals in the early stages of cognitive decline or mild dementia [39]. In line with evidence from Cochrane reviews and reports from the U.S. National Institute on Aging, which highlight CT as a promising non-pharmacological strategy to support cognitive performance in MCI [40–42], the program aimed to enhance performance in domains closely related to the study’s outcome measures, including global cognitive functioning, verbal episodic memory (immediate, learning, and delayed), and short-term verbal memory.

Training was administered at home using paper-based workbooks, designed for accessibility among older adults with limited digital skills. The program consisted of 10 daily sessions of approximately 60 min each, with progressively challenging tasks tailored to each participant’s cognitive level. Task difficulty was graded individually based on participants’ performance during training. Progression criteria included accuracy, task completion, and error rates. When participants consistently performed tasks with high accuracy, task difficulty was increased by adding stimulus load, reducing cues, or increasing task complexity. If participants showed difficulty, task demands were maintained or reduced to ensure engagement and feasibility. Exercises targeted 3 main domains: (a) verbal episodic memory, through tasks involving immediate and delayed recall, promoting encoding and consolidation strategies; (b) verbal learning, using repetition-based and semantic categorization tasks to strengthen acquisition; and (c) attention and short-term memory, via sequencing and span tasks analogous to forward digit recall.

Harms

Adverse events were systematically assessed at each intervention session using a standard tDCS adverse effects checklist (skin irritation, tingling, headache, dizziness).

Procedure

Participants were referred by clinicians from the Neurology Department of the Consorcio Hospital General Universitario of Valencia as potential candidates for the study. Following an initial telephone contact, an in-person appointment was scheduled in which the research procedures were explained in detail. After providing written informed consent, participants completed a demographic interview and underwent cognitive and functional assessments to verify eligibility.

Individuals who met all inclusion criteria were randomly assigned to one of four experimental groups using stratified block randomization by gender: (1) tDCS without CT, (2) tDCS with CT, (3) sham tDCS without CT, or (4) sham tDCS with CT.

The assigned intervention was delivered at the participant’s home over 10 consecutive days. Participants were required to attend all 10 intervention sessions; adherence to the full protocol was mandatory for trial participation. In the combined conditions, tDCS and CT were administered simultaneously during each session. Cognitive assessments were conducted at 3 time points: baseline (pre-intervention), post-intervention (immediately after the 10th session), and follow-up (1 month later). All cognitive assessments were conducted by evaluators who were not involved in the intervention delivery and were blinded to participants’ group allocation. Baseline assessments were performed prior to randomization, and post-intervention and follow-up assessments were conducted under the same blinded conditions.

Statistical analysis

A priori power analysis was conducted using G*Power 3.1 to estimate the required sample size for detecting a medium effect size (f = 0.25) in a 4 × 3 mixed ANOVA (group × time), with α = 0.05 and power (1-β) = 0.80. The analysis indicated a minimum of 36 participants. The final sample of 41 participants therefore met this requirement.

Statistical analyses were conducted using SPSS version 29. First, descriptive statistics and baseline homogeneity tests were performed using ANOVA for continuous variables and chi-square tests (χ²) for categorical variables. Participants who withdrew prior to completing the intervention were excluded from the analyses. A per-protocol analysis was therefore conducted, including only participants who completed the full intervention and all assessment time points. Consequently, no missing data were present in the analyzed dataset.

To evaluate the effects of the intervention, 4 × 3 mixed-design ANOVAs (group × time) with Bonferroni corrections for multiple comparisons were conducted. The assumptions of ANOVA were verified through graphical inspection of normality and homogeneity of variances, as well as Mauchly’s test of sphericity for within-subject factors. The significance level was set at p < .05. In addition, partial eta squared (η²ₚ) was calculated to estimate the magnitude of observed effects.

Within-group effect sizes (Cohen’s d) were calculated for baseline-to-post-intervention and post-intervention-to-follow-up changes using pooled standard deviations to facilitate clinical interpretation. Values of 0.20, 0.50, and 0.80 indicate small, medium, and large effects, respectively.

Results

Table 2 presents the main and interaction effects of the intervention on primary cognitive measures. Significant time × group interactions were observed for the MMSE, TAVEC Trial 1, TAVEC Total, and Forward Digit Span. These results indicate that the trajectory of cognitive performance over time differed significantly between groups for these variables. In contrast, TAVEC Delayed Recall did not show a significant interaction, although it did present a main effect of time, suggesting changes regardless of the intervention group.

Table 2.

Main and interaction effects on primary cognitive variables

Variable	Time Effect			Group Effect			Time × Group Interaction
	F(2, 74)	p	η²ₚ	F(3, 37)	p	η²ₚ	F(6, 74)	p	η²ₚ
MMSE	20.54	<.001	.357	3.63	.022	.227	6.72	<.001	.353
TAVEC trial 1	5.96	.004	.139	1.01	.400	.076	7.25	<.001	.370
TAVEC total	10.15	<.001	.215	3.90	.016	.240	6.35	<.001	.340
TAVEC delayed	8.16	.001	.181	2.28	.096	.156	1.30	.268	.095
Digit Span Forward	10.77	<.001	.225	3.67	.021	.229	4.19	<.001	.335

Table 3 summarizes the means and standard deviations of the variables that showed significant interactions, broken down by group and time of assessment.

Table 3.

Means, standard deviations, mean changes, and effect sizes for variables showing significant group × time interactions

Variable	Group	Baseline	Post-intervention	Follow-up	ΔP-BL	ΔFU-P	d(BL-P)	d(P-FU)
MMSE	tDCS + CT	25.18 (1.66)	26.36 (1.12)	27.55 (1.04)	1.18	1.19	0.83	1.10
	tDCS	25.09 (1.58)	26.64 (1.96)	26.82 (1.94)	1.55	0.18	0.87	0.09
	Sham + CT	24.89 (1.76)	26.44 (1.51)	26.44 (2.07)	1.55	0.00	0.95	0.00
	Sham	24.70 (2.06)	23.9 (2.23)	23.70 (3.34)	-0.80	-0.20	-0.37	-0.07
TAVEC trial 1	tDCS + CT	2.64 (0.67)	3.55 (0.82)	4.73 (0.91)	0.91	1.18	1.22	1.36
	tDCS	3.09 (1.22)	3.82 (1.25)	3.36 (1.29)	0.73	-0.46	0.59	-0.36
	Sham + CT	2.89 (1.16)	3.78 (1.64)	3.67 (1.41)	0.89	-0.11	0.63	-0.07
	Sham	3.50 (0.85)	2.50 (1.08)	2.80 (1.40)	-1.00	0.30	-1.03	0.24
TAVEC total	tDCS + CT	22.91 (2.07)	28.18 (4.94)	32.55 (6.49)	5.27	4.37	1.39	0.76
	tDCS	25.64 (6.97)	31.36 (8.45)	28.82 (10.69)	5.72	-2.54	0.74	-0.26
	Sham + CT	24.56 (9.67)	29.67 (9.26)	29.78 (10.60)	5.11	0.11	0.54	0.01
	Sham	21.60 (5.06)	20.70 (7.42)	16.50 (7.04)	-0.90	-4.20	-0.14	-0.58
Digit Span Forward	tDCS + CT	7.09 (1.87)	8.55 (1.86)	9.01 (1.54)	1.46	0.46	0.78	0.27
	tDCS	7.27 (1.27)	8.64 (1.86)	8.45 (1.44)	1.37	-0.19	0.86	-0.11
	Sham + CT	6.67 (1.01)	7.44 (1.01)	7.33 (1.41)	0.77	-0.11	0.76	-0.09
	Sham	6.80 (2.30)	6.20 (2.09)	5.90 (1.52)	-0.60	-0.30	-0.27	-0.16

Δ values represent mean changes scores between assessment points

Bl Base line, P Postintervention, FU Follow-up

Given the significant time × group interactions in the primary cognitive variables, post hoc analyses were performed to examine in greater detail the changes in cognitive performance for each measure.

For the MMSE, within-group post hoc analyses revealed that the three intervention groups (tDCS combined with CT, tDCS without CT, and sham with CT) showed significant improvements between baseline and post-intervention (p < .01 in all cases), as well as between baseline and follow-up (p < .05 in all cases). In addition, the group that received combined tDCS and CT maintained the improvement between post-test and follow-up (p = .027), suggesting a sustained progression in cognitive performance.

In the between-group analysis, no significant differences were found at baseline (p > .05), indicating adequate equivalence across conditions. After the intervention, the tDCS groups (with or without CT) and the sham tDCS with CT group obtained significantly higher scores than the sham tDCS without CT (p < .05). At follow-up, these differences were maintained only in the tDCS groups, particularly in the combined tDCS with CT condition.

For TAVEC Trial 1, within-group post hoc analyses revealed significant time effects across all 4 groups. The 3 groups including an intervention component (tDCS with or without CT, and sham tDCS with CT) showed significant improvements between baseline and post-intervention (p < .05 in all cases). Moreover, the group receiving combined tDCS and CT showed a sustained progression of verbal performance, with significant differences between baseline and follow-up (p < .001), and between post-intervention and follow-up (p = .018). In contrast, the sham without CT group exhibited a significant decline between baseline and post-intervention (p = .004), which remained stable at follow-up, indicating a persistent decrease in immediate memory performance.

In the between-group analysis, no significant differences were observed at baseline (p > .05). At post-intervention, a trend toward significance was detected (p = .066), though no pairwise comparisons reached significance. However, at follow-up, between-group differences were statistically significant, with the tDCS combined with CT group outperforming the sham tDCS without CT group (p = .007).

For the TAVEC Total score, within-group post hoc analyses revealed significant time effects across all four experimental groups. The 3 groups receiving an active intervention component (tDCS with CT, tDCS without CT, and sham tDCS with CT) showed significant improvements between baseline and post-intervention, and between baseline and follow-up (p < .05), except for the baseline-follow-up comparison in the tDCS without CT group, which did not reach statistical significance. The group receiving combined tDCS and CT showed a sustained progression of verbal performance, with significant differences between post-intervention and follow-up. In contrast, the sham tDCS without CT group showed a declining pattern, with a significant decrease between baseline and follow-up (p = .041).

In the between-group analysis, no significant differences were found at baseline (p > .05). After the intervention, significant differences emerged (p = .001), with the 3 groups receiving an active intervention component scoring significantly higher than the sham tDCS without CT group. This pattern persisted at follow-up (p = .002), with intervention groups maintaining superior performance compared with the sham tDCS without CT group, without significant differences between them.

For the Forward Digit Span, within-group post hoc analyses revealed significant time effects only in the groups receiving tDCS. Both the tDCS combined with CT group and the tDCS without CT group showed significant improvements between baseline and post-intervention (p < .001 in both cases), as well as between baseline and follow-up (p < .005). In both groups, performance remained stable between post-intervention and follow-up, indicating a sustained improvement over time. In contrast, the sham tDCS groups (with or without CT) did not show significant changes across the 3 assessments, suggesting a stable trajectory without intervention-related improvements.

In the between-group analysis, no significant baseline differences were observed (p > .05). After the intervention, significant differences were found between tDCS groups (with or without CT) and the sham tDCS without CT group (p < .05 in both cases), with the tDCS groups showing higher scores. These differences were maintained at follow-up (p < .001), with both tDCS groups outperforming the sham without CT group. No significant differences were found between the 2 tDCS groups or between them and the sham tDCS with CT group.

All participants completed the planned 10/10 sessions, yielding 100% adherence. No protocol deviations occurred. No adverse events were reported by participants during the trial.

Discussion

This study examined the efficacy of a home-delivered, non-pharmacological intervention combining tDCS and structured CT, two approaches previously investigated in isolation, in older adults with MCI-AD. The study was designed to compare the individual and combined effects of tDCS and CT across multiple cognitive domains. In line with our a priori hypothesis, the most pronounced improvements emerged in global cognition, immediate memory, verbal learning, and short-term verbal memory in the group receiving combined tDCS and CT. Importantly, this combined condition showed a more robust and sustained pattern of improvement over time compared with the other experimental conditions, maintaining gains at follow-up across all cognitive domains that showed significant effects.

Consistent with previous studies, our results demonstrated significant improvements in global cognitive functioning across all three intervention groups, with the most robust and sustained effect observed in the combined condition (tDCS + CT). This pattern suggests a moderate synergistic effect when both techniques are applied simultaneously, as in the present online protocol, reflecting a functional interaction between cortical stimulation and task-related activation. The temporal coincidence of these processes may have facilitated targeted modulation of relevant neural networks, in line with state-dependency theory, thereby supporting both the induction and consolidation of the cognitive changes observed [43].

The observed improvement is consistent with recent studies reporting positive effects of tDCS on global measures in individuals with MCI-AD or mild Alzheimer’s disease [23, 44], although it contrasts with broader reviews that have not found consistent benefits in this domain [45]. One possible explanation lies in the simultaneous combination of tDCS with active cognitive tasks, which may have enhanced neuronal responsiveness through activity-dependent plasticity mechanisms. In addition, the structured home-delivered setting may have supported adherence and minimized contextual factors that typically compromise cognitive performance, such as stress or demotivation. Interventions delivered outside the home may be associated with increased demotivation due to unfamiliar environments, travel burden, time constraints, and performance-related stress, whereas home-delivered interventions may help reduce these barriers and support sustained engagement. These conditions, together with the concurrent application of stimulation during functional activities, may have amplified the effects through state-dependency processes and targeted modulation of prefrontal networks.

With respect to immediate verbal memory, significant improvements were observed in all intervention groups, with a more pronounced and progressive effect in the combined condition (tDCS + CT), which sustained and further extended its gains at follow-up. This pattern suggests a possible synergistic effect when cortical stimulation coincides temporally with the active encoding of verbal information, facilitating the functional activation of frontotemporal networks involved in immediate memory.

These findings are consistent with prior studies reporting sustained improvements in immediate memory after tDCS [23], as well as evidence highlighting an enhancing effect when stimulation is combined with targeted cognitive tasks [46]. In this context, the coactivation induced by the dual intervention may have promoted functional plasticity within the relevant circuits, thereby improving the efficiency of early verbal information processing.

Finally, the decline observed in the sham tDCS without CT group, contrasted with the improvements observed in the tDCS groups and in the sham tDCS with CT group, further supports the specificity of the observed intervention effects.

For verbal learning, assessed through cumulative performance on the TAVEC (TAVEC Total score), significant improvements were detected in all intervention groups. However, only the combined tDCS + CT group maintained and further enhanced these gains at follow-up. This pattern indicates that the greatest benefits in verbal learning occur when cortical stimulation via tDCS is applied simultaneously with tasks requiring repeated verbal encoding, as in the multiple TAVEC trials. Such temporal coincidence may have facilitated the progressive consolidation of learning through reinforced activation of frontotemporal networks.

These results are consistent with prior findings demonstrating improvements in verbal learning following tDCS applied to prefrontal regions involved in consolidation processes [47], as well as with evidence supporting the efficacy of combined interventions in enhancing memory functions [48]. In the present study, the use of a structured CT program suggests that accessible interventions can yield clinically meaningful benefits, particularly when combined with concurrent and targeted tDCS. This finding complements studies where CT was effective for other functions but not for verbal learning [49] Given that their sample included patients with more advanced impairment and did not incorporate cortical stimulation, our results reinforce the importance of intervening at earlier stages with combined strategies. Overall, these data support the view that, when tDCS is applied online and temporally coupled with relevant functional processing, benefits in verbal learning may be maximized through enhancing efficiency of the underlying networks.

With respect to short-term memory, as assessed with the Forward Digit Span task, significant improvements were observed only in the groups receiving tDCS, and these effects were maintained at follow-up. This pattern indicates a specific impact of tDCS on functions related to the maintenance of information, which are particularly sensitive to the modulation of cortical excitability in the relevant regions.

These findings are consistent with prior studies reporting improvements in short-term memory and attention following tDCS with the anode positioned over the left DLPFC [44, 50]. In our study, the fact that CT alone did not yield significant improvements on this measure reinforces the differential role of tDCS in domains with high executive demands, possibly due to its capacity to modulate frontoparietal networks involved in attentional control. This evidence supports the potential of tDCS as an adjunctive tool for strengthening specific cognitive resources, particularly in the early stages of cognitive decline.

In contrast with the other functions assessed, long-term memory did not show a significant time × group interaction. Although a main effect of time was observed, improvements were general and not differential across intervention conditions, suggesting a parallel trajectory in all groups.

This pattern is consistent with evidence indicating that delayed memory is less responsive to brief interventions, particularly in populations with MCI-AD, where long-term consolidation processes are typically more compromised [51]. It is possible that the duration or intensity of the intervention was insufficient to induce sustained changes in these mechanisms, or that the training tasks did not directly engage strategies for delayed recall or strategic retrieval.

Clinical and theoretical implications

The home-delivered combination of tDCS and CT produced significant and sustained improvements in key cognitive functions, particularly in verbal memory and global cognition, with stronger effects than those obtained by either component alone. These findings indicate a moderately synergistic effect and highlight the value of accessible interventions in the early stages of cognitive decline.

From a clinical perspective, our results reinforce the potential of combined tDCS and CT as a safe, accessible, and effective strategy for individuals with MCI-AD, especially at prodromal stages where functional reserve remains considerable [52, 53]. The home-delivered implementation, by demonstrating positive effects comparable to those observed in controlled settings, supports the feasibility of delivering neurocognitive programs in the home environment and may contribute to improving accessibility of non-pharmacological interventions [54, 55].

From a theoretical standpoint, these findings provide additional support for the hypotheses previously advanced regarding the coactivation of brain networks and state-dependency as key mechanisms underlying the synergy between tDCS and CT [56, 57]. Specifically, the greater benefits observed in the combined group can be explained by the temporal coincidence between cortical stimulation and the functional activation induced by structured cognitive tasks, which promotes activity-dependent synaptic plasticity processes. As outlined in Hebbian models of plasticity, this simultaneity facilitates neuronal depolarization and NMDA channel opening, thereby promoting the consolidation of learning [58]. Likewise, the state-dependency framework posits that the effectiveness of stimulation depends on the functional state of the targeted brain network during the task: modulation is effective only when the network is actively engaged [57]. This complementary framework helps explain why the combined intervention proved more effective and durable: it optimized both the “how” (induction of plasticity) and the “when” (functional state most conducive to modulation).

Moreover, this study complements and extends previous research that reporting positive effects of these interventions when applied separately [13, 59], by confirming that their combination not only increases the magnitude of the effect but also appears to sustain it over time. This supports the development of integrative intervention models, in which CT is adapted to the individual’s cognitive profile while tDCS is delivered using standardized protocols [60].

Although the intervention was delivered in the participants’ homes, it should be noted that the protocol relied on staff-assisted administration rather than self-administration by participants or caregivers. This approach was intentionally chosen to ensure safety, accurate electrode placement, and strict adherence to stimulation parameters, which is particularly relevant in combined tDCS and CT protocols. In addition, the CT component required continuous monitoring and task adaptation based on individual performance, making fully unsupervised delivery impractical within the present design. Importantly, staff-led delivery also ensured procedural equivalence across the four experimental conditions, thereby strengthening internal validity and comparability between groups. Although this home-delivered, staff-assisted model differs from fully self-administered home-based interventions, it represents a necessary step to establish efficacy and optimize protocols before broader implementation strategies can be considered.

Limitations

Several limitations should be acknowledged. First, the sample size was modest, which limits statistical power and the examination of individual moderators or predictors of response. Although the study was adequately powered to detect medium-sized effects, replication in larger samples is needed to confirm the robustness of the findings. Second, the follow-up period was relatively short (1 month), which precludes firm conclusions regarding the long-term stability of the observed benefits. Third, the study employed a single-blind design, as intervention staff were necessarily aware of group allocation, which may introduce expectancy effects despite blinded outcome assessment.

In addition, although the intervention was delivered in participants’ homes, it required daily visits from trained research staff rather than self-administration by participants or caregivers. This delivery model may limit scalability and differs from fully self-administered home-based protocols reported in the literature. However, staff-led delivery was necessary to ensure correct electrode placement, participant safety, and consistent implementation of the combined intervention, particularly given the need to adapt CT tasks to individual performance and to maintain procedural equivalence across the four experimental conditions. Future studies should examine whether similar effects can be achieved using caregiver-assisted or fully self-administered protocols once optimal parameters have been established.

Finally, while statistically significant effects were observed, effect sizes were explicitly reported to facilitate interpretation beyond p-values, particularly given the neurodegenerative nature of the condition and the modest magnitude of change expected in this population.

Conclusion

Taken together, our findings support the use of home-delivered combined interventions as an accessible and effective strategy to modulate cognitive functions in individuals with MCI-AD. The online, simultaneous application of tDCS and CT demonstrated superior efficacy and durability compared with either intervention alone, in line with theoretical models emphasizing the importance of temporal coupling between stimulation and functional activation. These results further support the development of integrative intervention approaches, grounded in neuroplastic mechanisms, aimed at early-stage intervention in the course of the Alzheimer’s disease.

Abbreviations

MCI-AD

Mild cognitive impairment due to Alzheimer’s disease

NIA-AA

National Institute on Aging-Alzheimer’s Association

CT

Cognitive training

MCI

Mild cognitive impairment

tDCS

Transcranial direct current stimulation

DLPFC

Dorsolateral prefrontal cortex

GDS

Global Deterioration Scale

M@T

Memory Alteration Test

MMSE

Mini-Mental State Examination

Authors' contributions

J.C.M. and J.E. conceived and designed the study and drafted the manuscript. A.P. and J.G. conducted the statistical analyses and wrote the results. E.S. and J.C.M. interpreted the findings in the context of previous research and wrote the discussion. J.E., E.S., and I.M. contributed to participant recruitment, neuropsychological assessment, and the design and application of the tDCS intervention. All authors read and approved the final manuscript.

Funding

This work was supported by Ministerio de Ciencia, Innovación y Universidades (MICIU)/ Agencia Estatal de Investigación (AEI) /10.13039/501100011033 and by the “European Union” Grant [PID2022-136798OB-I00].

Data availability

De-identified individual participant data, data dictionary, analysis code (SPSS syntax), and study materials will be available upon reasonable request from the corresponding author for 5 years after publication, for non-commercial research purposes and subject to ethics approval when required.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the principles of the Declaration of Helsinki. The protocol was approved by the Human Research Ethics Committee of the University of Valencia (Approval ID: UV-REG.2601758). The trial was prospectively registered at ClinicalTrials.gov (Identifier: NCT06861231) on February 28, 2025. All participants provided written informed consent prior to participation, with the assurance that they could withdraw from the study at any time. This trial adhered to the CONSORT 2025 guidelines to ensure rigorous reporting and transparency.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.