JMIR Publications

ABSTRACT

Background:

Insomnia is prevalent and imposes a significant public health burden. Digital health interventions(DHIs) offer convenient and continuous digital management models, substantially improving the accessibility of sleep health management. Nonetheless, the comparative clinical efficacy and adherence across digital interventions with diverse functional architectures remain to be established.

Objective:

To evaluate the comparative efficacy and adherence of diverse digital health interventions(DHIs) for the treatment of insomnia, and to investigate the associations of delivery support modalities and intervention duration with treatment adherence.

Methods:

PubMed, Embase, PsycINFO, and the Cochrane Library were systematically searched from inception to July 8,2026. Randomised controlled trials(RCTs) evaluating DHIs for insomnia were eligible. Evidence networks were stratified into human-supported and fully automated sub-networks. Single-arm proportion meta-analysis was used to assess adherence rates, and multivariable meta-regression was used to examine associations with support modality and intervention duration. Risk of bias and certainty of evidence were assessed using RoB 2 and CINeMA, respectively.

Results:

A total of 138 RCTs involving 24523 participants were included. In the human-supported sub-network, all DHIs except wearable interventions were associated with statistically significant, large reductions in insomnia severity versus passive controls (SMD: -0.79 to -2.71). VR, dBBIs, and AI-driven interventions achieved the highest comparative rankings, whereas App-based CBT-I and Telehealth CBT-I showed comparatively precise and consistent estimates. In the fully automated sub-network, all DHIs except Multimodal CBT-I showed statistically significant, moderate-to-large reductions (SMD: -0.53 to -1.43).VR, wearable interventions, and Web-based CBT-I ranked highest;App-based CBT-I and Web-based CBT-I showed the most consistent evidence. The pooled overall adherence rate was 73.24% (95% CI: 68.48%–77.51%), with human-supported interventions exhibiting significantly higher completion rates than fully automated interventions (80.6% vs. 65.3%). Human support was significantly associated with higher adherence (OR = 2.26, 95% CI: 1.475–3.459). Conversely, each 4-week increase in intervention duration was associated with a 47.7% reduction in the odds of treatment completion (OR = 0.53, 95% CI: 0.34–0.82), with no significant interaction between support mode and duration (OR= 1.01, 95% CI: 0.401–2.561).

Conclusions:

In human-guided settings, dBBIs,App-based CBT-I,and Telehealth CBT-I, alongside Wearable interventions,App-based CBT-I,and Web-based CBT-I in fully automated settings, represent clinically mature and highly accessible options.Emerging modalities, including VR, wearable interventions，AI-driven interventions, showed high adherence as measured by treatment completion.Human support was associated with higher adherence, whereas longer protocols were associated with lower adherence.