Dementia, a common neurodegenerative disease, was getting more and more attention with the progress of the global population aging. According to the World Health Organization (WHO), there were approximately 55 million cases of dementia patients worldwide, and the number of dementia patients will continue to rise as the world population ages. It is estimated that by 2,050, the number of dementia patients will increase to 139 million (Serge Gauthier et al., 2022). In addition, the total estimated cost of dementia was $1.3 trillion by 2020, which was set to rise with dementia patients in 2030 (World Health Organization, 2021). These ever-increasing patients have brought a double burden on society and economy, while the increase of the estimated prevalence and incidence of dementia emphasized the necessity of effective treatment.

At present, there was a controversy about the pathogenesis of dementia, which had led to the failure to make breakthroughs in drug research on the etiological treatment at this stage. The current drug treatment mainly included two types of cholinesterase inhibitors and ionotropic glutamate receptor antagonists (Tisher and Salardini, 2019), which may cause side effects such as gastrointestinal discomfort, constipation, syncope, falls, arrhythmias, and extrapyramidal symptoms (Cummings et al., 2019), although they had specific improvement effects on patients’ clinical manifestations. In contrast, non-pharmacological therapies (NPTs), which aimed at improving dementia in the elderly, had attracted considerable attention due to their safe, relatively inexpensive, and scalable intervention. At present, the routine NPTS research on dementia patients showed that exercise therapy (ET), cognitive stimulation therapy (CST), and computerized cognitive training (CCT) indicate better treatment effects (Liang et al., 2019). And enriched environment (EE) and photobiomodulation (PBM), as new treatment modalities, have shown sound therapeutic effects in recent studies (Bourdon and Belmin, 2021; Salehpour et al., 2021). However, the efficacy of NPTs were controversial, because the results of individual studies vary widely depending on the training contents. Meanwhile, most of the above NPTs in previous studies have focused on comparing the effectiveness of single non-pharmacological intervention with conventional care in reducing cognitive impairment in patients with dementia. The lack of direct comparative studies of different NPTs leads to differences on which non-pharmacological interventions are most effective (Sikkes et al., 2021).

To tackle this problem, a network meta-analysis is well suited, because it facilitates comparisons of multiple pairs of interventions in one statistical model (Dias et al., 2018).

It’s considered that there were few systematic reviews or meta-analyses have pooled data of randomized controlled trials (RCTs) of dementia patients covering all above aspects, especially the novel non-pharmacological treatment approaches. There was no evidence in the literature to prove which interventions is the best for improving the cognitive function of dementia patients. Therefore, this study provided an optimal evidence-based basis for selecting non-pharmacological treatment options for dementia patients by comparing the magnitude effects of different non-pharmacological treatments on dementia cognition through the frequentist model of network meta-analysis (network meta-analysis).

This systematic review was performed according to the Cochrane Handbook for the Systematic Review of Interventions (Deeks et al., 2019) and according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Liberati et al., 2009). The review protocol was registered with PROSPERO (CRD42022363746).

As far as we know, from the science, from the perspective of frequency science, especially when combining traditional and more recent NPTs, previous studies have not proposed which intervention is the best way to treat cognitive function of dementia patients. Therefore, direct and indirect evidence were used to evaluate the relative effectiveness of different NPTs in cognitive interventions of dementia patients. According to the available data summarized in this study, the efficacy of the five NPTs were ranked from good to bad were: PBM, EE, ET, CCT, and CST. PBM, EE, ET, and CST show significant differences compared to CON (p < 0.05).

This study analyzed the improvement of cognitive function in patients with dementia by non-pharmacological treatments, which was consistent with the findings of a previous network meta-analysis (Wang et al., 2020). All NPTs had great potential to improve cognitive performance in patients with dementia, while CST and ET were shown to be the most beneficial interventions. Based on this, this study included additional new non-pharmacological treatment modalities and further evaluated their therapeutic effects compared with traditional non-pharmacological treatments such as CCT, CST, and ET. It was found that CST and ET had better therapeutic effects in improving cognitive function in patients. Notably, we found that PBM performed best in treating cognitive dysfunction in dementia patients by adding the latest non-pharmacological treatment modalities through frequency science perspective.

Photobiomodulation refers to a type of light therapy that utilizes that visible or near-infrared (NIR) light (600–1,100 nm) from lasers or light-emitting diodes (LEDs) stimulates or modulates various cellular and biological processes (Salehpour et al., 2018; Glass, 2021). Although most of the studies included in this article used red light or NIR at wavelengths from 650 to 1,080 nm for intracranial and intranasal irradiation methods, the treatment regimen for PBM was still largely dependent on patient-physician preference. Our findings were consistent with those reported in a previous systematic review (Zhu et al., 2022) that PBM was effective as a novel therapeutic approach to improve the level of cognitive function in dementia patients. This positive effect of PBM relied on four potential mechanisms (Hamblin, 2016), the basic ones involving photon absorption in the mitochondria (cytochrome c oxidase) (Hennessy and Hamblin, 2017), terminal enzymes in the electron transport chain, triggering downstream molecular and biochemical pathways in the mitochondrial respiratory chain, and exerting therapeutic effects (Lane, 2006; Karu, 2008; Hamblin, 2018); and adjust regional cerebral blood flow to increase perfusion levels (Chao, 2019; Salehpour et al., 2020; Baik et al., 2021); and open the light-mediated of calcium channels (Jung et al., 2019; Kim et al., 2020) and promote the activation of signaling mediators and transcription factors (Wu et al., 2022). In addition, PBM can reduce Aβ production and plaque formation by shifting amyloid precursor protein (APP) to non-amyloidogenic pathways (Zhang et al., 2020). These specific mechanisms of PBM are effective in improving mitochondrial function and increasing oxygen activity and ATP production, inhibiting aspects such as the downregulation of inflammation through inhibition of the NF-κB pathway. These aspects have a more significant role in improving cognitive function in patients with dementia. Moreover, the results of this study show that PBM has better efficacy in improving cognitive function methods in dementia patients compared to the other four non-pharmacological treatments. PBM therapy was a safe, non-invasive, non-thermal, and economical approach to improving cognitive function in patients with dementia while significantly reducing the pain of treatment, the adverse effects of treatment, and the financial burden on the family. In conclusion, PBM was a promising non-pharmacological option associated with cognitive improvement in patients with dementia. However, the optimal treatment regimen for different dementia severity and other modifying factors needs to be clarified to provide more precise individualized treatment plans in the future.

The rank probability of efficacy indicated that EE ranked second in effectiveness among the five different non-pharmacological interventions. This finding was similar to previous reports that EE was effective in improving cognitive function in patients with dementia (Bourdon and Belmin, 2021; Cutuli et al., 2022). EE was a non-invasive treatment that provides plasticity to the brain by combining cognitive training, such as memory and thinking, with dynamic stimulation, such as color, sound, and light, in an enriched environment (Figuracion and Lewis, 2021). A large number of animal studies have demonstrated the superiority and effectiveness of “Enriched environments” in improving cognitive functions in the brain (Nakano et al., 2020; Cordier et al., 2021). However, the network meta-analysis showed that there was no significant difference between EE and other non-pharmacological treatments, which may be due to the different study methods in the data pool. In the network meta-analyses, various interventions had slight interactions in the adjusted pooling and data comparison. Currently, there were few studies in this area of EE, and it’s a need to use more rigorous designs, more standard protocols, and more extensive studies to evaluate the effects of EE on improving cognitive function in patients with dementia.

The results of this study suggested that exercise therapy significantly improved the cognitive function of patients, which was consistent with previous meta-analyses that exploring the effects of exercise on dementia patients (Jia et al., 2019; López-Ortiz et al., 2021; Huang et al., 2022). In recent years, ET had been widely used as a low-risk and low-cost non-pharmacological treatment for patients with dementia. A large number of RCTs had reported the positive effects of exercise on cognitive function in patients with dementia (Henskens et al., 2018; Lamb et al., 2018; de Oliveira Silva et al., 2019). Exercise therapy improved cognitive performance mechanisms, such as increasing growth factors (Ruiz-González et al., 2021; Xue et al., 2022), modulating inflammatory cytokines (Hashiguchi et al., 2020; de Farias et al., 2021), alleviating oxidative stress (Hu et al., 2022), increasing cerebral blood flow (Lu et al., 2019; Tomoto et al., 2021), decreasing antibody concentrations (Giménez-Llort et al., 2013), and inhibiting tau phosphorylation from slowing the progression of dementia (Wang et al., 2021; Xu et al., 2022). However, these potential mechanisms have been proved to exist only in animal models, and some studies examining these mechanisms have not yet to prove their applicability to humans. This research included medium and long-term aerobic exercise, resistance training, physical and mental exercise, Tai Chi exercise and multi-component exercise. There was moderate heterogeneity among all RCTs on ET, which indicated that ET may have some variability in cognitive improvement due to the differences in exercise modality, intensity, frequency, and duration of study design. It was worth noting that recent studies had some differences in exploring the effects of different exercise modalities on cognitive improvement. Some studies have found that multi-component exercise could better improve cognitive dysfunction in patients (McDermott et al., 2019; Yang et al., 2021). Recent studies showed that resistance training appears to have the best therapeutic effect on improving cognitive function in patients (Huang et al., 2022). Therefore, future research on the process of exercise therapy for cognitive improvement needs to describe more specific exercise modalities and find more accurate ways to mitigate the process of dementia.

The results showed that there was no significant difference in the effect of CCT on the cognitive performance of patients with dementia, which was consistent with the findings of previous meta-analysis studies (Gates et al., 2019). Several studies found that cognitive training improves cognitive function in multiple cognitive domains in patients with mild cognitive impairment and dementia (Kanaan et al., 2014; Trebbastoni et al., 2018). CCT, a new cognitive training system that presents cognitive training tasks in a computer program, had a better effect on cognitive function (Bauer and Andringa, 2020). Despite the vital role of CCT in improving cognitive function, the current findings were not optimistic, which may be due to fewer included studies or the lower sensitivity of MMSE to cognitive function changes than other scales (Fu et al., 2017). All RCTs in this study used the MMSE as an assessment tool, which can’t accurately evaluates subtle changes in cognitive function (Weuve et al., 2006).

Cognitive stimulation therapy can improve cognition more effectively than controls, which was consistent with the previous meta-analysis that reported a more significant effect on cognitive function (Saragih et al., 2022). Using repetitive activities, especially tasks and games can help improve brain connectivity and generate new synapses and myelinated neural circuits, which was contributed to restoring or reorganizing neuronal structures behind cognitive function (Bryck and Fisher, 2012). Meanwhile, studies showed that CST has better clinical efficacy than drug therapy (Liang, 2019; Devita et al., 2021). It was noteworthy that CST did not show better efficacy compared to other non-pharmacological therapies in the present study.

Our study also had some limitations. First, the quality of the included studies was moderately heterogeneous due to the significant differences in treatment frequency, and treatment modality between different NPTs. Secondly, EE in our network only included a few studies, and there was less evidence-based evidence from studies of EE as a new non-pharmacological treatment modality, which may make the results biased. Thirdly, most of the included studies in this network meta-analysis compared non-pharmacological treatments with controls, while the number of actual head-to-head trials was relatively small, so comparative efficacy between interventions was often based on indirect comparisons. Fourthly, although we assessed the three assumptions of the network meta-analysis (homogeneity assumption, transferability assumption, and consistency assumption) to ensure their plausibility, there was moderate heterogeneity. Finally, our study did not analyze the safety of cognitive interventions because only four included studies described their adverse effects.

In conclusion, our network meta-analysis concluded that the best non-pharmacological treatment modality for patients with dementia was PBM, followed by EE, ET, and CST. However, the results should be interpreted with caution, considering the limitations of our meta-analysis described above and the insufficient number of studies in the existing literature. In the future, more multi-arm randomized controlled trials should be conducted to provide more direct evidence for the relative effectiveness of various non-pharmacological treatment.

The original contributions presented in this study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

FG played a guiding role in the manuscript writing and revision and data analysis. FW contributed to the conception or design of the work, the writing and revision of the manuscript, and the processing of the data. All authors contributed to the article and approved the submitted version.