The effectiveness of digital technology interventions for cognitive function in older adults: a systematic review and meta-analysis of randomized controlled trials

Abstract

Digital technology interventions (DTIs) are seen as promising interventions to prevent or delay cognitive decline in older adults, yet evidence from reviews is not conclusive. The aim of this study is to explore the effectiveness of DTIs in improving older adults’ cognitive function while taking study design and intervention characteristics as moderators. We searched the PubMed, Embase, CINAHL, PsycINFO, and Scopus databases up to May 26, 2023. Only randomized controlled trials examined the effects of DTIs on cognitive function were included in our study. Standardized mean difference (SMD) and 95% confidence interval for outcomes were applied in meta-analyses and subgroup analyses. A risk of bias assessment was also conducted. Overall, 23 eligible studies with a total sample size of 1454 participants were included. We found that DTIs significantly improved global cognitive function (SMD = 0.479), attention and processing speed (SMD = 0.488), executive function (SMD = 0.287), immediate recall (SMD = 0.266), and working memory (SMD = 0.307). Our subgroup analyses revealed that DTIs were more effective for cognitively impaired subjects, and DTIs with specific intervention characteristics, such as the inclusion of cognitive standard tasks, virtual reality-based interventions, specialized settings, professional guidance, low/medium-density training, > 24 sessions, and sessions lasting > 30 min, were more effective for different cognitive domains. This study supported the effectiveness of DTIs in improving cognitive function in older adults aged 60 years old and over, which may be influenced by study design and intervention characteristics. These findings have important implications for clinical dementia prevention and treatment strategies targeted at specific cognitive domains.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-024-01446-z.

Introduction

The World Health Organization has indicated that between 2015 and 2050, the proportion of older adults aged 60 and over will double, from 12 to 20% [1]. Age-related cognitive function decline in multiple functional domains (e.g., attention and processing speed, short and long-term memory, executive function) is associated with various adverse outcomes that severely influence the quality of life in older adults [2, 3]. Without timely and appropriate intervention, cognitive impairment or varying degrees of cognitive deterioration will irreversibly progress to neurodegenerative diseases (Mild cognitive impairment) and even dementia [4, 5]. Non-pharmacological therapies are the most common interventions to mitigate cognitive impairment or dementia [6]. As emerging non-pharmacological therapies, digital technology interventions (DTIs, e.g., smartphone apps, computer-assisted therapy, and wearable technologies) could not only achieve universal health coverage [7–9], but also have overcome quarantine barriers during the special crisis such as the COVID-19 pandemic [10]. Therefore, the popularity and effectiveness of DTIs that focus on improving the older adults’ cognitive function have attracted a lot of attention.

Although several reviews and meta-analyses have examined cognitive health implications of DTIs [11–18], the effectiveness of DTIs on cognitive function remains controversial [19–23]. Of those, some studies have reported the effectiveness of DTIs on cognitive function in older adults, including improvements in global cognitive function and memory [11, 12]. However, a randomized controlled trial (RCT) conducted among older adults with dementia has not found a significant effect of DTIs on global cognitive function when compared to a traditional control group [21]. Similarly, a recent study focusing on DTIs based on computer usage suggests limited improvement in cognitive function among healthy older adults [23]. These inconsistent findings may be attributed to variations in study design and intervention characteristics across different trials. A recent review in patients with post-stroke cognitive impairment highlighted that the effects of DTIs on cognitive function differed depending on the type of intervention, with virtual reality-based interventions showing distinct effects compared to non-virtual reality interventions [24]. Additionally, the effects of DTIs on cognition were found to vary based on the study design, including factors such as sample characteristics [25, 26] and control conditions [19, 26] employed. These findings highlight the need for further investigation to better understand the effectiveness of DTIs on cognitive function in older adults.

More importantly, existing meta-analysis studies mentioned above have focused on the effectiveness of only one type of DTIs (e.g., only virtual reality-based exergames [11, 12]) or on only one domain of cognitive function (e.g., only global cognition [13] or memory [12]). Moreover, present analyses have not sufficient consideration of study design and intervention characteristics. Therefore, the objectives of this study are the following: (1) estimating the effectiveness of DTIs on global cognition and different domains of cognition function, including attention and processing speed, executive function, immediate recall, delayed recall, and working memory; (2) examining the influence of study design (sample characteristics and control conditions) and intervention characteristics (intervention type, delivery mode, delivery formats, setting, intensity of training, number of sessions/modules and time per session). This study could shed light on the development of the most effective DTIs targeted at each domain of cognitive function, which provides advice for clinical dementia prevention and treatment.

Methods

Participants

Participants were older adults aged 60 years and over. No restrictions were imposed on the duration and severity of cognitively related disorders. The sample population was required to complete pre-intervention and post-intervention cognitive function assessments to reduce bias and avoid the effect of confounding factors on the results. The sample characteristics were categorized by baseline cognitive status (healthy subjects; any cognitively impaired subjects; the whole population) using the criteria adopted by Millán-Calenti et al. [27].

Interventions

According to Shah et al. [28], the term DTIs refers to an approach that utilizes digital technology. This encompasses various technological tools, devices, and applications that process information in the form of numeric codes, typically binary code. Intervention characteristics included intervention type (single-domain cognitive standard tasks, multi-domain cognitive standard tasks, cognitive task-based games, others), delivery mode (non-virtual reality, virtual reality), delivery formats (unclear, guided), setting (home community, specialized setting, e.g., nursing home/hospital/clinic/daycare center), intensity of training (low-intensity, medium-intensity, high-intensity, unclear) [29], number of sessions/modules (1–8, 9–16, 17–24, > 24) [30], and time per session (≤ 30-min, > 30-min, unclear) [31]. The intervention types of single-domain cognitive standard tasks and multi-domain cognitive standard tasks refer to guided practice on a set of standardized tasks designed to reflect specific cognitive functions, such as memory, attention, language, and executive function [32]. Cognitive task-based games are interactive games that are specifically designed to target and enhance specific cognitive domains.

Control group

Any control conditions including no treatment, treatment-as-usual, and any other psychological or physical treatment were included.

Study design and outcomes

To reduce bias and to avoid the effect of confounding factors on results, only RCTs were considered [33] where the primary or secondary outcome was a validated measurement of cognitive function. RCTs could provide evidence of causation and are recognized as the gold standard for clinical trials [34].

Cognitive function was estimated at baseline and endpoint. If studies reported outcomes at more than one time point throughout the intervention, data from the longest follow-up period were selected based on prior literature [35]. Meanwhile, when multiple scales or tests measured the same cognitive domain in a study, the primary, gold standard, or more discriminatory and generalizable outcomes were adopted. Neuropsychological scales or tests were classified according to the domains of cognition, similar to previous reviews [35, 36]. Global cognitive function scales (e.g., the Mini-Mental State Examination) were considered and the domains included attention and processing speed (sustained alertness), executive function (a set of cognitive processes responsible for the planning and monitoring of goal-directed behaviors), immediate recall, delayed recall, and working memory (the short-term storage and operation of information).

Search strategy

Five online academic databases were selected, including PubMed (from 1984), Embase (from 2008), CINAHL (from 2006), PsycINFO (from 1991), and Scopus (from 1995). These databases were searched for publications focusing on the effect of DTIs on the cognitive function in older adults. Publications up to May 26, 2023, limited to English language, were included in the search. The search strategy was developed and adapted to fit the requirements for each of the databases. A combination of MeSH and free text terms for search terms with the following search string was used in database searches: ((Alzheimer* OR dementia* OR cognition OR cognitive function OR cognitive impairment OR cognitive decline OR executive function) AND (technol* OR digital OR online OR internet OR mobile OR electronic OR social media OR smartphone OR web-based OR telemedicine OR telehealth OR app OR virtual reality OR augmented reality OR AR OR wearable OR computer OR PDA OR EDA OR laptop OR e-reader OR Ehealth OR Elearning OR e-health OR (playable AND device) OR gamif* OR application) AND (intervention* OR program* OR training OR support* OR treatment* OR therapy)). In addition, references of identified studies were screened for potentially relevant resources. The full search strategy is available in supplementary eTable 1.

Study inclusion/exclusion criteria

Publications were included if they met the following inclusion criteria: (a) RCTs of DTIs for cognition/dementia through a digital platform (e.g., web-based/online/internet, electronic/smartphones/computers/laptops/tablet/e-readers/Ehealth/Elearning, social media/app, telemedicine/telehealth, wearable or playable device/application/gamification/virtual or augmented reality headsets) in older adults aged 60 years old and over; (b) having appropriate control group types, such as approaches based on non-DTIs or online course training that does not cover cognitive areas; (c) peer-reviewed studies; (d) full-text papers available in English.

Data extraction and risk of bias assessment

All identified studies were downloaded into EndNote X9, a reference management software. First, duplicates were removed. Then, three reviewers (CC, MYZ, and BH) independently screened all study titles and abstracts and compared the full text against the eligibility criteria. Any disagreements between two reviewers were discussed and resolved through consensus with a third reviewer.

The data extraction process commenced on June 30, 2023, and was carried out using a standardized template as outlined below: (a) author (publication year); (b) mean age and feature of the study sample; (c) outcome measured; (d) assessment tools of cognitive function; (e) control conditions and study duration; (f) intervention characteristic, including the mode of delivery, intensity of training, time per session, number of sessions/modules, delivery formats, and settings. All information was extracted from the review papers considered.

In order to ensure reliability of results, we used the revised Cochrane risk of bias tool for randomized trials (RoB 2, version 2) to access the risk of bias of each study [33]. The risk of bias assessment was also independently conducted by two reviewers (MYZ and BH), and disagreements were resolved by negotiation with a third reviewer (CC).

Data analysis

All statistical analysis was conducted using STATA 15.0 SE software. P-values < 0.05 were considered statistically significant.

To determine the pooled effect size of DTIs on cognitive function in older adults, we utilized the mean changes and standard deviations (SDs) of cognitive function scores to calculate the standardized mean difference (SMD) [35]. The SMD was calculated by the changes between the intervention group and control group, and the changes included both the mean terms and the SDs terms. We estimated the overall effect size by employing meta-analyses with random or fixed-effects models based on statistical heterogeneity. The formula of SMD was as below:

$${\text{Mean}}_{\text{change}}={\text{Mean}}_{\text{post}-\text{intervention}}-{\text{Mean}}_{\text{baseline}};{\text{SD}}_{\text{change}}=\text{square root}\left[{\left({\text{SD}}_{\text{baseline}}\right)}^2,+,{\left({\text{SD}}_{\text{post}-\text{intervention}}\right)}^2,-,\left(2,,\ast ,\text{Corr},,\ast ,{\text{SD}}_{\text{baseline}},,\ast ,{\text{SD}}_{\text{post}-\text{intervention}}\right)\right],$$

where Corr = 0.5, in accordance with the Cochrane Handbook 5.0.2 and previous studies [35, 37]. In cases where standard deviations were not provided, SDs were calculated from standard errors using a formula recommended in the Cochrane guidelines [38].

To examine the potential variability of the effect of DTIs on cognitive function across different study designs and intervention characteristics, we initially employed the I² statistic to assess heterogeneity among the included studies. Heterogeneity was considered significant if the corresponding P-value was equal to or less than 0.05, or if I² was equal to or greater than 50%. To explore the sources of heterogeneity, we conducted subgroup analyses based on the following criteria: (a) study design, including sample characteristics and control conditions; (b) intervention characteristics.

This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA, supplementary material eTable 2) guidelines. The study protocol was registered with the Prospective Register of Systematic Reviews (PROSPERO) protocol ID CRD42023430938.

Results

Search results

A total of 1800 study titles and abstracts were reviewed, and 192 full-text studies were identified for eligibility. All full-text studies were written in English. After excluding 162 studies that failed to meet the inclusion criteria among these full-text studies, 23 studies were included in the systematic review. Out of them, two records were excluded based on the following reasons: one record was specific to females [21] and another record lacked clear data annotation [39]. Finally, 21 were included in the meta-analysis (Fig. 1).

Fig. 1.

Preferred reporting items for systematic reviews and meta-analysis flow diagram of each stage of the study selection

Descriptive analysis of systematic reviews

Table 1 provides a summary of the 23 studies included.

Table 1.

Characteristics of studies included in the systematic review and meta-analysis

Author (The year of publication)	Sample characteristics	Cognitive scales/tests used	Control conditions	Intervention	Intervention types	Delivery mode	Intensity of training	Intervention time per session	Number of sessions/ modules	Delivery formats	Setting
Kim et al. (2021)	Healthy subjects	Modified Alzheimer’s Disease Assessment Scale–Cognitive Subscale, Korean Mini-Mental State Examination, Digit Span Test, Digit Symbol Coding, Korean Color–Word Stroop Test, Subjective Evaluation of Cognitive Function	No treatment	An communication technology (ICT)-based multicomponent program	Games involving cognitive tasks	Virtual reality training	Low-intensity (60 min/week)	30-min	24	Guided	Hospital
Kazazi et al. (2021)	Healthy subjects	Mini-Mental State Examination, the Wisconsin card sorting test, the Stroop test, the N back test, and the go/no go test	Educational workshops	Attentive Rehabilitation of Attention and the Memory (ARAM) software application	Multi-domain cognitive standard tasks	Non-virtual reality training	Low-intensity (90 min/week)	45-min	60	Unclear	Health centers
Park et al. (2022)	Cognitively impaired subjects	The Korean version of the Executive Function Performance Test	No description	Virtual shopping training	Others: virtual shopping training	Virtual reality training	Unclear	Unclear	16	Guided	Local senior centers
Park et al. (2020)	Cognitively impaired subjects	The Weschsler Adult Intelligence Scale-Revised Block Design Test, the Seoul Verbal Learning Test	No description	Virtual reality-based spatial cognitive training (VR-SCT)	Multi-domain cognitive standard tasks	Virtual reality training	Medium-intensity (135 min/week)	45-min	24	Guided	Local senior centers
Recio-Rodríguez (2022)	Healthy subjects	Mini-Mental State Examination, Clock test score, animal naming	Treatment-as-usual	A smartphone (Samsung Galaxy J3) and a smartband (Xiaomi Miband S2) on loan with the EVIDENT 3 app pre-installed	Others: lifestyle and physical activity advice	Non-virtual reality training	Low-intensity (25 min/week)	25-min	2	Unclear	Health centers
Ta´rraga et al. (2006)	Cognitively impaired subjects	Mini-Mental State Examination, The story recall subtest from the Rivermead Behavioral Memory Test, Syndrom Kurztest	Treatment-as-usual	An interactive multimedia internet-based system (IMIS)	Multi-domain cognitive standard tasks	Non-virtual reality training	Low-intensity (60 min/week)	20-min	36	Unclear	Daycare center
Torpil et al. (2021)	Cognitively impaired subjects	The Loewenstein Occupational Therapy Cognitive Assessment-Geriatric	only the cognitive rehabilitation	A virtual reality (computer-generated interactive environments) intervention in addition to a conventional cognitive rehabilitation intervention	Games involving cognitive tasks	Virtual reality training	Low-intensity (90 min/week)	45-min	24	Guided	Occupational therapy department
Vanoh et al. (2019)	Cognitively impaired subjects	Mini-Mental State Examination, digit symbol, Rey Auditory Verbal Learning Test, visual reproduction I and II,matrix reasoning, Clock Drawing Test, Trail Making Test-A	Health education pamphlets: the ‘Healthy Eating’ pamphlet	A web-based application, WESIHAT 2.0©	Others: Ten lifestyle-based memory enhancing guides	Non-virtual reality training	Medium-intensity (120 min/week)	30-min	96	Unclear	Community
Wahbeh et al. (2016)	Cognitively impaired subjects	Auditory-Verbal Learning Test, Controlled Oral Word Association, Flanker Task, Letter-Number Sequencing, Simple Reaction-Time Task	Educational session Control	Internet Mindfulness Meditation Intervention	Others: mindfulness courses learning	Non-virtual reality training	Low-intensity (60 min/week)	30-min	12	Guided	Community
Zhang et al. (2022)	Healthy subjects	Digit Symbol Substitution, Letter Set, Shipley Institute of Living Scale, Stroop Color Name Test, Trail Making Test-A and B, Perception of Memory Functioning	Parallel content delivered in a non-digital format	A computer system: PRISM system	Others: social support program	Non-virtual reality training	Unclear	Unclear	7	Unclear	Community
Zhuang et al. (2013)	Cognitively impaired subjects	The Addenbrooke’s Cognitive Examination-Revised scale	No treatment	A human–computer interaction-based comprehensive cognitive training	Games involving cognitive tasks	Non-virtual reality training	High-intensity (225 min/week)	75-min	72	Unclear	Nursing home
Zukowski et al. (2022)	No specific status	Clock task	Conventional treadmill training (CTT)	A virtual reality treadmill training (VRTT)	Others: exercise training	Virtual reality training	Low-intensity (30 min/week)	30-min	1	Unclear	Community
Eckroth-Bucher et al. (2009)	Subjects in both states	Mini-Mental State Examination, the Dementia Rating Scale, the Wechsler Memory Scale – III	Treatment-as-usual	The Integrated Cognitive Stimulation and Training Program (ICSTP)	Multi-domain cognitive standard tasks	Non-virtual reality training	Low-intensity (90 min/week)	15-min	36	Guided	Community
Diaz Baquero et al. (2022)	Cognitively impaired subjects	Mini-Mental State Examination, the cognitive subscale of the Alzheimer’s Disease Assessment Scale, Trail Making Test-A and B, the Clock Drawing Test, Digit Symbol, WAIS-III, Visual Memory of the Rivermead Behavioural Memory Test, Visual Reasoning of the Cambridge Cognitive Examination, and Verbal Fluency Test	Treatment-as-usual	Computer-based cognitive training programs— GRADIOR	Multi-domain cognitive standard tasks	Non-virtual reality training	Low-intensity (60–90 min/week)	30-min	7	Guided	Hospitals/day centers
İnel Manav et al. (2019)	Cognitively impaired subjects	Standardized Mini-Mental State Examination	Unstructured casual conversations	Internet-based videos	Single-domain cognitive standard tasks	Non-virtual reality training	Low-intensity (60 min/week)	60-min	12	Guided	Nursing home
Kang et al. (2021)	Cognitively impaired subjects	Mini-Mental State Examination, Rey-Osterrieth Complex Figure Test, copy task, the digit span forward and backward test and Trail Making Test-A, the Korean version of the Boston Naming Test, the Seoul Verbal Learning Test, Trail Making Test-B, and the Stroop Color Test	Treatment-as-usual	Multi-domain and neuropsychologist-assisted cognitive training in a fully immersive virtual reality environment	Multi-domain cognitive standard tasks	Virtual reality training	Low-intensity (40–60 min/week)	20–30-min	8	Guided	Clinic
Menengi C et al. (2022)	Cognitively impaired subjects	Mini-Mental State Examination	No treatment	Exercise treatment via home-based telerehabilitation	Others: exercise treatment	Non-virtual reality training	Low-intensity (25 min/week)	15–40-min	25	Guided	Home
Millán-Calenti et al. (2015)	Healthy subjects	Mini-Mental State Examination	No treatment	A computerized cognitive training application	Multi-domain cognitive standard tasks	Non-virtual reality training	Unclear	20-min	24	Unclear	Community
Nousia et al. (2021)	Cognitively impaired subjects	Montreal Cognitive Assessment, Clock Drawing Test, immediate word recall, word recognition and delayed word memory test, Boston Naming Test, Semantic Fluency measure, the digit span forward and digit span backward test, Trail Making Test-A and B	Treatment-as-usual	A computer-based multi-domain cognitive training program with the use of the RehaCom software	Multi-domain cognitive standard tasks	Non-virtual reality training	Medium-intensity (120 min/week)	60-min	30	Unclear	Clinical laboratory
Nousia et al. (2018)	Cognitively impaired subjects	Montreal Cognitive Assessment, Trail Making Test A and B, the digit span forward; digit span backward, repeat and word recognition and delayed memory, verbal fluency, Boston Naming Test	Treatment-as-usual	Multi-domain Cognitive Training (MCT) program	Multi-domain cognitive standard tasks	Non-virtual reality training	Medium-intensity (120 min/week)	60-min	30	Guided	Community
Moon et al. (2020)	Cognitively impaired subjects	Mini-mental status examination for dementia screening	Storytelling with no digital materials	Digital reminiscence therapy app on the Android platform	Cognitive task training	Non-virtual reality training	Low-intensity (60 min/week)	30-min	8	Guided	Daycare centers
Luo et al. (2023)	Cognitively impaired subjects	Montreal Cognitive Assessment and other cognitive measures	HE: cognitive training courses	rEAP: synchronous video teaching of art-themed activities	Single-domain cognitive standard tasks	Non-virtual reality training	Medium-intensity (120 min/week)	60-min	24	Guided	Community
Oliveira et al. (2021)	Cognitively impaired subjects	Frontal Assessment Battery, the Trail Making Test, Mini-Mental State Examination and the Clock Drawing Test	Treatment-as-usual	A virtual reality cognitive stimulation at residential care homes	Multi-domain cognitive standard tasks	Virtual reality training	Low-intensity (90 min/week)	45-min	16	Guided	Residential care homes

Population

Overall, 1454 subjects (737 in the digital intervention group and 717 in the control group) were included into the meta-analysis. A total of five studies (21.7%) were conducted in healthy subjects [23, 25, 27, 40, 41], 16 studies (69.6%) included cognitively impaired subjects [19–21, 26, 39, 42–52], and one study (4.3%) enrolled subjects in both statuses [32]. Only one study (4.3%) lacked a detailed description of the specific cognitive function status [53].

Intervention characteristics

The types of DTIs included in the studies were broadly categorized into four groups: single-domain cognitive standard tasks (n = 2) [21, 43], multi-domain cognitive standard tasks (n = 9) [26, 27, 32, 40, 42, 47–50], cognitive task-based games (n = 4) [41, 44, 51, 52], and other types (n = 8) [19, 20, 23, 25, 39, 45, 46, 53]. Approximately half of the 23 eligible studies (n = 11, 47.8%) utilized cognitive standard tasks to intervene in the cognitive function of older adults. Non-virtual reality training (n = 16, 69.6%) [19–21, 23, 25–27, 40, 42, 43, 45–48, 52, 54] was the most commonly used mode of digital intervention delivery. Specialized settings (n = 15, 65.22%) [21, 25, 26, 39–45, 47, 49–52] were the most frequently employed environment for subject recruitment and intervention implementation. Among the intervention formats, 14 studies (60.9%) [19, 21, 32, 39, 41–46, 48–51] clearly stated the inclusion of a professional therapist or supervisor to provide guidance. Regarding the curriculum, the majority of studies employed low-density training (n = 14, 60.9%) [20, 21, 25, 26, 32, 40–44, 46, 49, 51, 53], shorter duration for the courses (n = 11, 47.8%) [19–21, 25–27, 32, 41, 42, 44, 53], and a greater number of courses or modules (> 24) (n = 8, 34.8%) [19, 26, 32, 40, 46–48, 52].

Control conditions

Most DTIs studies included treatment-as-usual (n = 8) [25, 26, 32, 42, 44, 47–49] or any other psychological or physical treatments (n = 9) as control conditions; four studies considered no treatment control conditions [27, 41, 46, 52]. Among the nine studies that included other treatments, three considered educational programs [19, 20, 40], two considered cognitive training programs [45, 51], and one considered a physical training program [53]. Additionally, parallel content (n = 1) [23] and storytelling (n = 1) [21] delivered in a non-digital format and unstructured casual conversations (n = 1) [43] were also taken into consideration.

Outcome measures

A majority of studies (n = 18, 78.3%) involved global cognitive function rather than only specific domains of cognition. Scales that assessed global cognitive function include the Mini-Mental State Examination [19, 21, 25–27, 32, 40–44, 46, 49], Montreal Cognitive Assessment [45, 47, 48], the Loewenstein Occupational Therapy Cognitive Assessment-Geriatric [51], and Addenbrooke’s Cognitive Examination-Revised scale [52]. The most frequent global cognitive function measure was the Mini-Mental State Examination (13 records, 56.5%). However, three of these 18 studies lacked clearer data on baseline and post-intervention cognitive function assessments [47, 48, 54], and one adopted a short version of the Mini-Mental State Examination for specifically dementia screening [21]. Therefore, 14 studies were included in the final meta-analysis for global cognitive function.

More details are available in eTable 3 (Supplementary material).

Effectiveness of digital technology interventions on cognitive function

Global cognitive function

For global cognitive function (containing 14 studies), this study encompassed a total of 669 participants, with 338 subjects assigned to the intervention group and 331 to the control group. DTIs significantly improved global cognitive function (SMD = 0.479, 95% confidence interval (CI) 0.140 to 0.818; P < 0.01). Notably, there was considerable heterogeneity observed (I² = 76.0%; P < 0.001) (Fig. 2 (a)).

Fig. 2.

Forest plot for the effect of digital technology intervention on cognitive function, expressed as standardized mean differences between intervention and control groups. (a) Global cognitive function. (b) Attention and processing speed. (c) Executive function. (d) Immediate recall. (e) Delayed recall. (f) Working memory

Attention and processing speed

In relation to the attention and processing speed, a total of 891 participants were included in 14 studies, with 448 subjects in the intervention group and 443 in the control group. The results indicated a significant enhancement in attention and processing speed through the utilization of DTIs (SMD = 0.488, 95%CI 0.191 to 0.785; P < 0.01). Similarly, substantial heterogeneity was present (I² = 75.5%; P < 0.001) (Fig. 2 (b)).

Executive function

This study incorporated 12 studies involving 900 participants, with 455 subjects in the intervention group and 445 in the control group, to explore the impact on executive function. DTIs significantly improved executive function (SMD = 0.287, 95%CI 0.060 to 0.515; P < 0.05). Again, significant heterogeneity was observed (I² = 75.5%; P < 0.001) (Fig. 2 (c)).

Immediate recall

In terms of immediate recall, this study included 11 different studies. In total, there were 485 participants involved, with 249 subjects assigned to the intervention group and 236 to the control group. DTIs showed a significant improvement in immediate recall, compared to the control group (SMD = 0.266, 95%CI 0.085 to 0.447; P < 0.01). Across the 11 studies, there was an insignificant level of heterogeneity (I² = 37.9%, P = 0.097) (Fig. 2 (d)).

Delayed recall

For the delayed recall, this analysis included 7 studies, with a total of 308 participants. Among them, 159 subjects were part of the intervention group, while 149 were assigned to the control group. The results did not show any significant effect between the control and intervention groups (SMD = 0.491, 95%CI − 0.177 to 1.159; P > 0.05). A meaningful heterogeneity was observed across the 7 studies (I² = 86.9%, P < 0.001) (Fig. 2 (e)).

Working memory

Regarding working memory, which was examined in 7 studies, a total of 485 participants were included. Among them, 249 subjects were assigned to the intervention group, while 236 were assigned to the control group. The results indicated a significant enhancement in immediate recall among participants who received DTIs, in comparison to those in the control group (SMD = 0.307, 95%CI 0.073 to 0.542; P < 0.01). It is important to note that across the 7 studies, there was no significant heterogeneity observed (I² = 13.0%, P = 0.330) (Fig. 2 (f)).

Subgroup analysis

To determine the heterogeneity between studies, results of the subgroup analyses of global cognitive function, attention and processing speed, executive function, and delayed recall are demonstrated in Tables 2 and 3.

Table 2.

The subgroup analyses by the study design for the pooled standardized mean differences in the global cognitive function and other domains

Variables	Subgroup analysis
	Number of studies	SMD (95% CI)	I2 (P-value) (%)
Global cognitive function
Sample characteristics
-Healthy subjects	4	0.174 (− 0.166, 0.514)	31.5 (0.223)
-Cognitively impaired subjects	10	0.600 (0.134, 1.065)	79.8 (< 0.001)
-The whole population	-
Control conditions
-No treatment	3	0.475 (0.005, 0.944)	0 (0.698)
-Treatment-as-usual	7	0.251 (− 0.186, 0.688)	74.0 (0.001)
-Any other treatment	4	0.949 (0.078, 1.820)	87.9 (< 0.001)
Attention and processing speed
Sample characteristics
-Healthy subjects	3	0.204 (− 0.119, 0.528)	33.6 (0.222)
-Cognitively impaired subjects	11	0.561 (0.187, 0.935)	75.7 (< 0.001)
-The whole population	-
Control conditions
-No treatment	1	− 0.157 (− 0.848, 0.535)	-
-Treatment-as-usual	7	0.669 (0.096, 1.241)	82.6 (< 0.001)
-Any other treatment	6	0.356 (0.094, 0.617)	42.2 (0.124)
Executive function
Sample characteristics
-Healthy subjects	4	0.199 (− 0.153, 0.551)	66.8 (0.029)
-Cognitively impaired subjects	8	0.348 (0.029, 0.666)	54.3(0.032)
-The whole population	-	-	-
Control conditions
-No treatment	1	− 0.036 (− 0.726, 0.654)	-
-Treatment-as-usual	6	0.324 (− 0.112, 0.760)	73.7 (0.002)
-Any other treatment	5	0.248 (0.009, 0.486)	25.3 (0.253)
Delayed recall
Sample characteristics
-Healthy subjects	-	-	-
-Cognitively impaired subjects	6	0.549 (− 0.224, 1.322)	88.8 (< 0.001)
-The whole population	1	0.133 (− 0.563, 0.828)	-
Control conditions
-No treatment	-	-	-
-Treatment-as-usual	4	1.004 (0.096, 1.912)	86.5 (< 0.001)
-Any other treatment	3	− 0.148 (− 0.693, 0.398)	55.4 (0.106)

Table 3.

The subgroup analyses by the intervention characteristics for the pooled standardized mean differences in the global cognitive function and other domains

Variables	Subgroup analysis
	Number of studies	SMD (95% CI)	I2 (P-value) (%)
Global cognitive function
Intervention type
-Single-domain cognitive standard tasks	1	3.137 (2.083, 4.191)	-
-Multi-domain cognitive standard tasks	5	0.327 (0.024, 0.629)	0 (0.700)
-Games involving cognitive tasks	4	0.371 (− 0.411, 1.154)	82.6 (0.001)
-Others	4	0.192 (− 0.099, 0.483)	28.7 (0.240)
Mode of delivery
-Non-virtual reality	10	0.508 (0.127, 0.889)	75.1 (< 0.001)
-Virtual reality	4	0.369 (− 0.474, 1.211)	82.5 (0.001)
Delivery formats
-Unclear	6	0.202 (− 0.020, 0.423)	308 (0.392)
-Guided	8	0.647 (0.036, 1.258)	84.5 (< 0.001)
Setting
-Community	2	0.373 (− 0.101, 0.846)	0 (0.651)
-Nursing home/hospital	12	0.498 (0.108, 0.889)	58.83 (< 0.001)
Intensity of training
-Low-intensity	10	0.538 (0.057, 1.018)	83.1 (< 0.001)
-Medium-intensity	2	0.312 (− 0.043, 0.668)	0 (0.972)
-High-intensity	1	0.270 (− 0.424, 0.963)	−
-Unclear	1	0.752 (− 0.158, 1.662)	−
Number of sessions/modules
−1–8	3	− 0.075 (− 0.323, 0.174)	0 (0.830)
−9–16	2	1.685 (− 1.147, 4.516)	93.6 (< 0.001)
−17–24	4	0.584 (− 0.046, 1.214)	74.0 (0.009)
- > 24	5	0.430 (0.138, 0.722)	0 (0.908)
Intervention time per session
- ≤ 30-min	7	0.054 (− 0.149, 0.256)	0 (0.616)
- > 30-min	6	0.920 (0.243, 1.597)	84.2 (< 0.001)
-unclear	1	0.752 (− 0.158, 1.662)	−
Attention and processing speed
Intervention type
-Single-domain cognitive standard tasks	-
-Multi-domain cognitive standard tasks	6	0.516 (0.175, 0.858)	50.9 (0.070)
-Games involving cognitive tasks	4	0.725 (− 0.310, 1.759)	89.2 (< 0.001)
-Others	4	0.158 (− 0.030, 0.345)	0 (0.583)
Mode of delivery
-Non-virtual reality	10	0.291 (0.072,0.509)	40.3 (0.089)
-Virtual reality	4	0.972 (0.175,1.769)	83.9 (< 0.001)
Delivery formats
-Unclear	6	0.178 (0.004, 0.352)	0 (0.432)
-Guided	8	0.683 (0.191, 1.174)	80.2 (< 0.001)
Setting
-Community	4	0.429 (− 0.093, 0.950)	74 (0.009)
-Nursing home/hospital	10	0.511 (0.123, 0.899)	76.1 (< 0.001)
Intensity of training
-Low-intensity	7	0.571 (− 0.007, 1.149)	81 (< 0.001)
-Medium-intensity	5	0.577 (0.248, 0.905)	44.6 (0.125)
-High-intensity	1	− 0.157 (− 0.848, 0.535)	−
-Unclear	1	0.085 (− 0.141, 0.311)	−
Number of sessions/modules
−1–8	3	0.262 (− 0.165,0.690)	61.8 (0.073)
−9–16	1	− 0.008 (− 0.988, 0.972)	−
−17–24	4	0.816 (− 0.031, 1.662)	88 (< 0.001)
- > 24	6	0.441 (0.096, 0.785)	47.5 (0.090)
Intervention time per session
- ≤ 30-min	6	0.301 (0.028, 0.574)	2 (0.404)
- > 30-min	7	0.721 (0.213, 1.230)	81.8 (< 0.001)
-Unclear	1	0.085 (− 0.141, 0.311)	−
Executive function
Intervention type
-Single-domain cognitive standard tasks	-
-Multi-domain cognitive standard tasks	4	0.679 (0.110, 1.247)	74.0 (0.009)
-Games involving cognitive tasks	3	− 0.094 (− 0.489, 0.302)	0 (0.371)
-Others	5	0.171 (0.010, 0.332)	0 (0.872)
Mode of delivery
-Non-virtual reality	10	0.352 (0.113,0.592)	58.2 (0.011)
-Virtual reality	2	− 0.160 (− 0.845,0.525)	48.5 (0.163)
Delivery formats
-Unclear	6	0.297 (0.068, 0.527)	39.9 (0.139)
-Guided	6	0.227 (− 0.256, 0.711)	71.4 (0.004)
Setting
-Community	4	0.450 (− 0.127, 1.028)	78.7 (0.003)
-Nursing home/hospital	8	0.230 (− 0.018, 0.477)	40.4 (0.109)
Intensity of training
-Low-intensity	6	0.178 (− 0.138, 0.494)	44.4 (0.109)
-Medium-intensity	4	0.608 (0.123, 1.094)	67.4 (0.027)
-High-intensity	1	− 0.036 (− 0.726, 0.654)	−
-Unclear	1	0.084 (− 0.142, 0.311)	−
Number of sessions/modules
−1–8	4	0.131 (− 0.036, 0.299)	0 (0.685)
−9–16	1	0.134 (− 0.847, 1.115)	−
−17–24	2	− 0.096 (− 0.853, 0.661)	65.6 (0.088)
- > 24	5	0.621 (0.184, 1.058)	62.0 (0.032)
Intervention time per session
- ≤ 30-min	6	0.143 (− 0.070, 0.356)	0 (0.470)
- > 30-min	5	0.602 (0.158, 1.045)	65.9 (0.020)
-Unclear	1	0.084 (− 0.142, 0.311)	−
Delayed recall
Intervention type
-Single-domain cognitive standard tasks	-
-Multi-domain cognitive standard tasks	3	1.254 (0.167, 2.341)	86.8 (0.001)
-Games involving cognitive tasks	1	0.272 (− 0.348, 0.892)	−
-Others	3	− 0.148 (− 0.693, 0.398)	55.4 (0.106)
Mode of delivery
-Non-virtual reality	6	0.527 (− 0.266,1.320)	89.0 (< 0.001)
-Virtual reality	1	0.272 (− 0.348,0.892)	−
Delivery formats
-Unclear	2	0.692 (− 1.540, 2.924)	96.0 (< 0.001)
-Guided	5	0.426 (− 0.243, 1.095)	80.8 (< 0.001)
Setting
-Community	4	0.260 (− 0.829, 1.349)	89.6 (< 0.001)
-Nursing home/hospital	3	0.770 (− 0.164, 1.705)	86.8 (0.001)
Intensity of training
-Low-intensity	3	0.084 (− 0.336, 0.504)	0 (0.472)
-Medium-intensity	4	0.846 (− 0.215, 1.906)	92.4 (< 0.001)
-High-intensity	-
-Unclear	-
Number of sessions/modules
−1–8	1	0.272 (− 0.348, 0.892)	−
−9–16	1	− 0.503 (− 1.500, 0.495)	−
−17–24	1	0.273 (− 0.189, 0.734)	−
- > 24	4	0.821 (− 0.349, 1.992)	92.3 (< 0.001)
Intervention time per session
- ≤ 30-min	4	− 0.102 (− 0.488, 0.284)	21.7 (0.281)
- > 30-min	3	1.275 (0.164, 2.358)	90.3 (< 0.001)
-Unclear

Subgroup analysis categorized by study design

In terms of study design, the results are described in Table 2. DTIs demonstrated a significant improvement in the global cognitive function (SMD = 0.600, 95%CI 0.134 to 1.065), attention and processing speed (SMD = 0.561, 95%CI 0.187 to 0.935), and executive function (SMD = 0.348, 95%CI 0.029 to 0.666) among individuals with cognitive impairment, except for delayed recall. With respect to the control conditions, studies utilizing no treatment as the control group exhibited a significant effect of DTIs on global cognitive function (SMD = 0.475, 95% CI 0.005 to 0.944). Moreover, findings indicated a significant influence of DTIs on attention and processing speed (SMD = 0.669, 95% CI 0.096 to 1.241) as well as delayed recall (SMD = 1.004, 95% CI 0.096 to 1.912) compared with treatment-as-usual as the control group. Furthermore, when compared to other psychological or physical treatments, DTIs exhibited significant improvement in cognitive function, demonstrating a preference for the intervention in terms of global cognitive function (SMD = 0.949, 95% CI 0.078 to 1.820), attention and processing speed (SMD = 0.356, 95% CI 0.094 to 0.617), and executive function (SMD = 0.248, 95% CI 0.009 to 0.486).

Subgroup analysis categorized by intervention characteristics

Table 3 provides a summary of the results of subgroup analysis categorized by intervention characteristics. DTIs with multi-domain cognitive standardized tasks and sessions lasting more than 30 min were most effective in improving attention and processing speed, executive function, and delayed recall. Notably, a significant improvement in delayed recall was observed following the completion of multi-domain cognitive standardized tasks (SMD = 1.254, 95%CI 0.167 to 2.341) and interventions lasting more than 30 min per session (SMD = 1.275, 95%CI 0.164 to 2.358). In terms of delivery mode, virtual reality interventions resulted in a greater enhancement of attention and processing speed (SMD = 0.972, 95%CI 0.175 to 1.769) compared to non-virtual reality interventions (SMD = 0.291, 95%CI 0.072 to 0.509). Moreover, interventions conducted in specialized settings and professional guidance group yielded significant improvements in global cognitive function and attention and processing speed. The low/medium-density training group and groups consisting of more than 24 sessions demonstrated significant SMDs across all cognitive function domains, except for delayed recall.

Publication bias

Funnel plots (Fig. 3) and Egger’s tests suggested that no statistical significant publication bias existed in studies in all cognitive domains, including global cognitive function (t = 1.99, P = 0.070), attention and processing speed (t = 1.34, P = 0.205), executive function (t = 0.650, P = 0.528), immediate recall (t = 0.890, P = 0.396), delayed recall (t = 0.27, P = 0.797), and working memory (t =  − 0.13, P = 0.911).

Fig. 3.

Funnel plot of included dependent effect sizes

Risk of bias

The risk of bias assessment of the included study is summarized in Fig. 4. Many of the included studies had substantial drop-out rates, which resulted in the missing data. The studies lacked detailed information on the blinding procedure and allocation concealment, which were judged as a “High” risk of bias assessment [25, 40, 43, 44, 49, 52]. Three studies were evaluated as having “Some concerns” regarding the risk of bias considering they explicitly stated that participants were not blinded, although efforts were made to blind other staff members as much as possible [41, 47].

Fig. 4.

Risk of bias of included studies

Discussion

Summary of main findings

In the present analysis, to determine whether DTIs were effective in improving cognitive function in older adults aged 60 years old and over, we pooled the findings of 23 available peer-reviewed RCTs for systematic review and 21 available peer-reviewed RCTs for meta-analysis. The systematic review provided a narrative summary for qualitative synthesis and a meta-analysis for statistical synthesis of the findings. In our review, 19 of 23 studies showed significant improvements in at least one cognitive domain in the intervention groups compared with the control group at follow-up. In our meta-analysis, our findings suggested that DTIs significantly improved the global cognitive function as well as its four domains including attention and processing speed, executive function, immediate recall, and working memory among older adults. Moreover, the effect of DTIs on cognitive function differed depending on study design and intervention characteristics.

Comparisons with similar researches

Our main analysis found that DTIs had significant positive effects on cognitive function and its other domains except delayed recall, without publication bias. First, DTIs considerably raised the level of global cognitive function in older adults. This result is consistent with previous meta-analysis studies regarding various types of DTIs such as games [31, 55, 56]. Second, DTIs could improve attention and processing speed, executive function, immediate recall, and working memory but not delayed recall performance. It may be because delayed recall was the most impaired domain in older adults [57]. Although the pooled effect size (SMD) was not statistically significant, the results also indicated that delayed recall level of old adults is higher in the DTIs group than in the control groups. Besides, a previous study pointed out that the delay recall in the subjects of the DTIs group may be improved over time [58], which may influence the outcome.

Study design including sample characteristics and control conditions could influence the effectiveness of DTIs on cognitive function in older adults. Regarding sample characteristics, the finding that DTIs may produce larger improvements in global function for individuals with cognitively impaired than other groups is novel and warrants further investigation. A previous piece of evidence from a positron emission tomography scan may potentially support this result, which indicated that patients with cognitively impaired retain cognitive plasticity and even have a greater activation in the brain regions than healthy older adults when stimulated by a memory-related cognitive function training task [59]. Later, the results of pathologic studies of biopsy specimens from cognitively impaired patients showed an increase in synaptic contact size [26]. Thus, this suggested that interventions may be some utility to patients with cognitively impaired. Additionally, the effect of DTIs on delayed recall, executive function, and attention and processing speed was higher in studies which used treatment-as-usual as controls than in studies which used any other psychological or physical treatment as controls. This was because other psychological or physical treatments may have more active components conferring cognitive benefits that may be absent in usual therapy [60].

Intervention characteristics, including intervention types, formats, mode of delivery, setting, and other session information, may influence the cognitive benefits of DTIs. Regarding intervention types, most studies utilized single- or multi-domain cognitive standard tasks, which specifically targeted cognitive function improvement [61]. Consistent with previous meta-analyses on computerized cognitive training, our study found that digital cognitive standard task training had significant positive effects on global cognitive function and various domains such as attention and processing speed, executive function, and delayed recall [29]. It should be noted, however, that the effects of DTIs with other types of intervention, such as social support programs or exercise treatments, on executive function and delayed recall were not conclusive due to limited study numbers. Nevertheless, a social support program may foster old adults’ social connectedness through regular contact with professionals and peers [62]. Compared to DTIs with cognitive task-based games, DTIs with cognitive standard task training exhibited greater cognitive benefits. This difference can be attributed to the comprehensive treatment of each cognitive domain and increased interaction between older adults and instructors in cognitive standard task training. Meanwhile, cognitive task-based games tend to focus on a single cognitive area, lack specificity, and may not effectively achieve desired goals within a limited time frame, thus yielding smaller cognitive benefits. As for the mode of delivery, only a few studies utilized virtual reality technologies. Consequently, there was insufficient evidence to suggest a substantial impact of using virtual reality on overall cognitive function, executive function, and delayed recall. It is worth noting, however, that the utilization of virtual reality in DTIs yielded more positive outcomes in attention and processing speed when compared to DTIs without virtual reality. This observation could be attributed to the enhanced focus, decreased stress levels, and improved decision-making abilities that virtual reality environments offer [63–66]. As a result, older adults may experience psychological and social advantages, leading to better interactive skills and cognitive rehabilitation [63–66]. In our study, the findings suggest that cognitive function may be improved when participants receive intervention guidance from a professional therapist in a specialized setting, which is characterized by a higher level of organization and professionalism [67]. Additionally, studies involving online rehabilitation programs have also reported more favorable outcomes for groups that completed interventions in a guided format [68]. Detailed session information revealed that low- and medium-density training, a higher number of sessions, and longer session durations might bring greater cognitive functioning benefits. These findings can be explained by the longer memory consolidation time and lower memory burden associated with longer sessions, which are especially beneficial for older adults.

Possible underlying mechanisms of digital technology interventions

DTIs included in this study involved cognitive task/curriculum training, active motor game training, and social support programs. When older adults are given new courses, multicomponent and cognitive complexity provide them with more opportunities for creativity, which may promote self-exploration and enhance processing and cooperation between different cognitive domains, and thus improve cognitive functioning. Meanwhile, targeted cognitive training was beneficial for specifically improving the corresponding domains of cognitive function [61], which enables the brain to produce new synapses through neuroregeneration. Previous studies have shown that cognitive plasticity and learning potential apply to both Alzheimer’s disease patients and healthy older adults. Education [69] and psychomotor stimulation [70] are able to modulate the risk of cognitive deterioration and to supplement with additional neural resources in the brain, compensating for age-related cognition changes [71, 72]. The basic principle that active exergames training are generally beneficial for enhancing cognitive function is the experience-dependent neural plasticity [73], such as progressively increased task difficulty, task-oriented exercises, and transfer effects [74]. The so-called transfer effects refers to transferring improved function in a specific function to a different domain [75]. Therefore, the type of training games in DTIs may provide a transfer effect from motor to cognitive skills in older adults, similar to previous studies [76–79].

When the DTI was delivered by virtual reality, an additional spatial feature could be added with visual and auditory stimuli [63] for older adults [64, 65]. The additional environment may improve concentration, reduces stress, and supports decision-making to gain psychological and social benefits of interactive skills, thereby promoting cognitive rehabilitation in older adults [64–66, 80]. Besides, DTIs are able to treat patients online in a familiar environment via courses learning or specific domains training. DTIs online have shown comparable outcomes to traditional face-to-face therapy procedures [81], with cognitive functional areas containing articulation disturbances [82], aphasia [83], and vocabulary retrieval [84, 85]. Patients with treatment in a familiar environment have less dependency and mortality than patients receiving conventional treatment modalities [86–88], as they are able to reintegrate into their living environment earlier and have a better quality of life. In addition, guidance provided during DTIs, was considered to be important in assisting participants and maintaining adherence [89]. Given that the interventions with professional therapist/supervisor yielded positive health outcomes in the present study, it will be needed for future research to conduct larger trials to assess the benefit of including guidance with medical professionals.

Recommendations for future research

According to the results obtained from this review, it is advisable to incorporate additional studies of superior quality in order to facilitate more comprehensive evaluations of the effectiveness of cognitive function improvement among older adults who have been exposed to various forms of DTIs.

Specific recommendations for future studies are summarized as follows:

1. Include more RCTs with low risk of bias and a longer follow-up;

2. Collect functional magnetic resonance imaging or diffusion tensor imaging to track brain changes after DTIs interventions;

3. Identify a standardized objective cognitive measurement to estimate the effect of DTIs on different domains of cognition;

4. Pay attention to whether and how different DTIs improve delayed recall;

5. Repeatedly conduct RCT research using active controls as a control group to further verify if control conditions moderate the effect of DTIs on cognition and to control the placebo effects.

Limitations and strengths

There are several limitations in this review. First, RCTs included in this review varied considerably in their selection of cognitive assessment tools, blinded assessment methods, and intention-to-treat analyses; some of the included RCTs had low quality. Second, only two studies had a long follow-up period of more than 12 months [23, 42]; thus, we could not estimate the long-term effectiveness of DTIs on cognition. Third, most studies adopted scales/tests to assess cognitive function; there is a lack of clinical consensus on measure of various domains of cognitive function. And the applications of different DTIs may cause activation in different brain regions that are closely related to cognition, but such objective assessments of cognitive function have less been used. Fourth, limited studies included in our meta-analyses examine the effect of DTIs on delayed recall; we should be cautious about the conclusion that there was no significant positive effect of DTIs on delayed recall. Fifth, few included studies used no intervention or any other active psychological or physical treatments as controls. And active controls could be helpful to control placebo effects resulting from the intervention group [90].

Despite these limitations, it is important to acknowledge several notable strengths of this study. Firstly, this research represents the most comprehensive systematic review to date, incorporating meta-analyses and focusing extensively on the efficacy and feasibility of DTIs in older adults across various cognitive domains within RCTs. The methods employed were rigorous, with careful assessment of study quality and consideration of potential publication bias. Additionally, the inclusion of subgroup analyses examining study and intervention characteristics was a key strength, facilitating exploration of heterogeneity between studies and offering insights for the prevention and treatment of cognitive impairment in clinical settings. Subsequently, this approach enables the identification of the most effective DTIs tailored to specific populations.

Conclusions and practical implications

Overall, this review furnishes robust evidence supporting the efficacy of DTIs in improving cognitive function among older adults, specifically in attention and processing speed, executive function, immediate recall, and working memory. Nevertheless, it is important to consider that study design and intervention characteristics may influence the observed effects. These findings have significant clinical implications, validating the rationale behind the use of DTIs to enhance cognitive function. Moreover, by considering the nuances of study design and intervention characteristics, a more comprehensive understanding of the potential benefits and limitations of DTIs in older adults’ cognitive function can be achieved. As there are limited studies using long-term follow-up interventions, whether the conclusions of this study are applicable to explain the long-term effectiveness of DTIs needs further elucidation. Consequently, future research should prioritize addressing variations in study design and intervention characteristics to establish more definitive conclusions. Considering the broader implications, the potential impact of implementing such programs on a large scale cannot be underestimated. It could lead to improvements in overall health and cost reduction. Policymakers and clinicians should therefore actively promote the development and implementation of universal, indicated, and selective DTIs that specifically target the cognitive function of older adults in nursing homes and communities.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

Study concept and design: J.G., C.C.; data selection and data collection: C.C., N.H., B.H., M.Z.; data analysis: C.C.; drafting of the manuscript: C.C., N.H., J.Y.; critical revision of the manuscript for important intellectual content: J.G., N.H., and C.C.; C.C. and N.H. contributed equally as co-first authors. All authors agreed on the final manuscript and the decision to submit it for publication.

Funding

This study was supported by the Beijing Municipal Natural Science Foundation (grant number: L242145) and National Natural Science Foundation of China [grant number: 22376006 and 82071552]. The founders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

This paper reports meta-analytic data based on original published studies. The corresponding author, Jing Guo, could be contacted for further information.

Declarations

Conflict of interest

The authors declare no competing interests.