Journal of Medical Internet Research

Introduction

Obesity stays a crucial world well being problem, with far-reaching implications for morbidity, mortality, and well being care programs. According to the newest knowledge from the World Health Organization (WHO), the worldwide prevalence of weight problems amongst adults aged 18 years and older has greater than doubled, rising from 7% in 1990 to 16% in 2022 []. Recent proof from 2023 additional signifies that if present traits persist, over 4 billion individuals, greater than half of the world’s inhabitants, might be residing with obese or weight problems by 2035, inserting an unprecedented financial burden on world well being care infrastructures []. This escalating pattern is projected to exert a staggering financial affect, with the worldwide financial burden estimated to achieve US $4.32 trillion yearly by 2035, a determine corresponding to the affect of the COVID-19 pandemic []. Overweight or weight problems is a significant threat issue for quite a few noncommunicable ailments—together with heart problems, sort 2 diabetes, musculoskeletal issues, and sure cancers—and contributes considerably to the worldwide burden of illness and untimely dying []. This rising epidemic necessitates accessible, sustainable, and scalable methods for weight administration.

Lifestyle interventions, which usually combine dietary modification, bodily exercise enhancement, and behavioral change strategies, have emerged as cornerstone methods within the scientific administration of obese or weight problems []. Traditionally, these interventions have been delivered via face-to-face counseling or group-based periods and have demonstrated efficacy in lowering physique weight and enhancing cardiometabolic outcomes []. Structured behavioral parts, akin to objective setting, self-monitoring, and drawback fixing, are well known and integrated into scientific follow tips for obese or weight problems remedy [].

However, a key limitation of conventional life-style interventions is their reliance on in-person supply, which poses appreciable boundaries to entry. The requirement for bodily attendance could hinder participation amongst people in rural or underserved communities, these with mobility or scheduling constraints, and people going through socioeconomic disadvantages []. These accessibility challenges had been additional exacerbated through the COVID-19 pandemic, which disrupted in-person care worldwide and underscored the pressing want for scalable alternate options []. In the postpandemic period, there was a major shift towards digital-first well being care fashions, pushed by the necessity for cost-effective options that may bypass the constraints of human-led medical sources [].

Digital well being applied sciences provide a compelling answer to those limitations. Digital life-style interventions (DLSIs)—together with web-based platforms, cellular apps, and SMS textual content messaging–primarily based programs—allow the supply of customized, interactive, and cost-effective behavioral help at scale []. These interventions can promote engagement via real-time suggestions, self-monitoring instruments, and versatile entry, making them notably well-suited for huge implementation throughout numerous populations []. Recent developments in synthetic intelligence (AI) and automatic tailoring have additional enhanced the potential of those platforms to offer high-quality, customized help with out the necessity for fixed human supervision [,].

Prior systematic opinions have reported that digital interventions yield modest advantages over typical offline approaches in facilitating weight reduction [,]. However, these conclusions are sometimes constrained by the inclusion of hybrid interventions that mix digital parts with adjunctive modalities, akin to web-based platforms supplemented by therapist help or face-to-face counseling. Such designs make it tough to isolate the unbiased impact of digital platforms. Further complicating the proof base, one meta-analysis included on-line interventions delivered in group codecs—akin to digital group conferences—which, whereas not involving in-person contact, could introduce behavioral confounders, akin to peer accountability and social facilitation []. Additionally, one other evaluate targeted completely on web-based interventions, excluding various digital modalities akin to cellular apps or SMS textual content messaging–primarily based instruments, and thereby didn’t seize the broader spectrum of recent digital supply programs [].

Given the constraints of prior proof syntheses—lots of which have included hybrid or group-mediated interventions—there stays a crucial want for a targeted meta-analysis that isolates the unbiased affect of digital supply. While hybrid fashions integrating in-person teaching have proven promise, their excessive operational prices and lack of scalability restrict their utility for large-scale public well being applications []. This creates a crucial data hole concerning the stand-alone efficacy of digital know-how. By isolating digital interventions from individualized human-facilitated parts, this systematic evaluate and meta-analysis gives a clearer estimation of digital intervention results. This methodological refinement eliminates confounding from human-facilitated parts, permitting for a clearer evaluation of the know-how’s intrinsic affect. Furthermore, by synthesizing knowledge throughout a variety of supply codecs, together with cellular apps, internet platforms, and SMS textual content messaging–primarily based programs, this examine captures the modern panorama of digital well being instruments. This methodological focus permits for a extra exact and high-certainty evaluation of the unbiased results of stand-alone DLSIs on weight administration in adults with obese or weight problems.

Therefore, the first goal of this examine was to guage the efficacy of stand-alone DLSIs on anthropometric and dietary outcomes in adults with obese or weight problems. Specifically, we aimed to (1) quantify the pooled impact dimension of those interventions in comparison with varied management circumstances, (2) establish potential moderators of effectiveness via subgroup analyses, and (3) assess the knowledge of the proof to offer scientific and coverage suggestions for scalable weight administration.

Methods

Overview

This systematic evaluate and meta-analysis follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) tips and was prospectively registered with PROSPERO (CRD420251053974; ). While the examine was performed in response to the preregistered protocol, we moreover used the Hartung-Knapp-Sidik-Jonkman (HKSJ) methodology for the random-effects mannequin and calculated 95% prediction intervals (PIs) to offer a extra conservative and sturdy estimation of the proof, which weren’t explicitly detailed within the preliminary protocol.

Eligibility Criteria

Studies had been included in the event that they met the next standards: (1) adults (≥18 years) with obese or weight problems (BMI≥23 kg/m²); (2) the intervention concerned life-style interventions geared toward weight administration, delivered completely by way of digital platforms or applied sciences, together with cellular apps, pc software program, web sites, or SMS textual content messaging; (3) the intervention was delivered independently with none in-person human help. We outlined stand-alone DLSIs as these the place the core therapeutic parts—together with suggestions, objective setting, and progress monitoring—had been pushed completely by digital platforms. Specifically, we included all digital interventions that didn’t require bodily, face-to-face scientific contact, akin to automated or periodic app notifications, SMS textual content messaging reminders, and preprogrammed, asynchronous suggestions (eg, batch-sent academic emails or system-generated responses triggered by user-entered knowledge), as these signify scalable alternate options to conventional resource-intensive applications. For research utilizing probably ambiguous terminology (eg, “coaching,” “messaging,” or “peer support”), we verified the non–in-person nature of those parts by cross-referencing examine protocols or technical implementation sections. This verification ensured that any human involvement was neither in-person nor one-to-one help; (4) a non-DLSI management group was used for comparability, akin to lively controls, nonspecific lively controls, or waitlist controls; (5) outcomes included anthropometric measures (eg, physique weight, BMI, or waist circumference) or dietary measures (eg, caloric consumption); and (6) the examine design was a randomized managed trial (RCT). Exclusion standards had been as follows: (1) individuals weren’t primarily chosen primarily based on obese or weight problems standing; (2) individuals had been pregnant ladies; (3) any intervention incorporating in-person human help (eg, face-to-face counseling or bodily scientific visits) the place the unbiased impact of the digital know-how couldn’t be remoted; (4) interventions involving resource-intensive, one-to-one digital interplay with human suppliers. Specifically, we excluded labor-intensive, real-time interplay akin to customized telephone calls or particular person video conferencing, as these don’t align with the scalable nature of stand-alone DLSIs; (5) research included DLSIs within the management group; and (6) research missing ample knowledge for meta-analysis had been excluded. For probably eligible research with incomplete or ambiguous knowledge, we tried to contact the corresponding authors to acquire the lacking info. If no response was acquired or the supplied knowledge remained insufficient for calculating impact sizes, the research had been excluded from each the systematic evaluate and the meta-analysis. Both peer-reviewed papers and convention abstracts had been included in the principle evaluation, though convention abstracts had been required to report RCT findings explicitly.

Information Sources

The digital databases had been searched by the corresponding writer utilizing MEDLINE, Embase, PsycINFO, Web of Science, and the Cochrane Library to seek for peer-reviewed papers in English, with no begin date and no language restrictions. To guarantee the particular indexing and search capabilities of every database had been totally used, all databases had been searched individually via their respective official interfaces reasonably than utilizing a multidatabase search platform. We didn’t search examine registries (eg, the WHO International Clinical Trials Registry Platform) for unpublished or ongoing trials, as our search was targeted on peer-reviewed, printed literature.

Search Strategy

We performed and reported our literature search in accordance with the PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analyses–Search Extension) tips to make sure transparency and reproducibility [] (). The search methods had been developed independently for this examine and didn’t undertake or reuse search strings from prior literature opinions. The methods had been constructed by combining 3 primary domains (digital interventions, weight problems or obese, and weight administration) utilizing a complete set of Medical Subject Headings and related key phrases tailor-made to every database’s indexing construction []. To guarantee transparency and reduce the danger of lacking research, we adopted the Cochrane Handbook tips for search technique growth. The full, unabridged search technique for every database is supplied in . However, the search technique didn’t bear a proper peer evaluate course of by an exterior librarian.

The reference lists of all included research and related research had been searched manually, and the grey literature was additionally searched by getting into the identical key phrases into Google on May 18, 2025. During the preparation and revision of the meta-analysis, the search was repeated 4 occasions to establish newly printed trials. The final search was performed on March 4, 2026, following the earlier replace on June 30, 2025.

Selection Process

After eliminating duplicate data, a 2-phase screening process was carried out: preliminary screening of titles and abstracts, adopted by a full-text evaluate. Two unbiased reviewers (SAL and JHP) assessed the eligibility of every report at each levels utilizing Covidence (Veritas Health Innovation). To keep consistency throughout the method, a random 10% subset of entries was collectively reviewed throughout every screening section to calculate interrater settlement. Any disagreements had been resolved via dialogue and consensus between the two reviewers till a mutual settlement was reached. Interrater reliability, quantified utilizing Cohen kappa, indicated near-perfect settlement at each the title and summary screening stage (0.92) and full-text evaluate (0.94). Reviewers remained blinded to one another’s assessments all through to make sure objectivity. After finishing the literature search, 2 unbiased reviewers screened the titles and abstracts of all retrieved data to evaluate eligibility. Full-text papers had been subsequently obtained for trials deemed probably related and reviewed intimately for inclusion. In addition, the reference lists of all included papers and associated evaluate papers had been manually examined to establish any supplementary research.

Data Collection

Data had been extracted from every included examine by 2 unbiased reviewers (SAL and JHP) utilizing a standardized, pilot-tested knowledge extraction kind to make sure accuracy and consistency. Any discrepancies within the extracted knowledge had been resolved via detailed dialogue and inner cross-checking between the two reviewers. If important knowledge had been lacking or unclear within the unique experiences, we tried to contact the first authors by way of e mail to acquire the required info.

Data Items

From every included examine, the next info was extracted when obtainable: (1) bibliographical knowledge (ie, first writer and 12 months); (2) pattern traits (ie, pattern dimension, imply age, and inhabitants); (3) intervention traits (ie, particulars of intervention and management circumstances, goal life-style, theoretical framework, supply format, period, and final result domains); (4) participant retention charges, outlined because the proportion of individuals within the intervention group who accomplished the postintervention evaluation relative to the entire variety of individuals initially randomized; and (5) knowledge required to calculate within-group or between-group impact sizes for RCTs. To calculate impact sizes, each baseline (preintervention) and postintervention knowledge had been extracted for all outcomes. However, follow-up knowledge had been excluded from the meta-analysis to take care of consistency throughout research and keep away from bias ensuing from various follow-up durations.

Eligible interventions had been outlined as these aiming to alter weight-reduction plan and/or bodily exercise behaviors utilizing established habits change strategies (BCTs; eg, objective setting and self-monitoring). Interventions had been included if at the very least 50% of their content material focused habits change in both area and integrated a number of BCTs listed within the BCT Taxonomy v1 []. Detailed BCT classification for every included examine is supplied in . The main final result was anthropometric measures, together with physique weight, BMI, and waist circumference. When a examine reported a number of anthropometric variables, we utilized a predefined hierarchy to pick out a single final result for the first evaluation: (1) physique weight, (2) BMI, and (3) waist circumference. Secondary outcomes included dietary measures, with complete vitality consumption (kcal/d) prioritized as the first dietary metric.

Study Risk of Bias Assessment

Risk of bias in particular person research was appraised independently by 2 unbiased reviewers utilizing the Cochrane Collaboration’s Risk of Bias 2 device tips in 5 domains: randomization course of, deviations from meant interventions, lacking final result knowledge, measurement of the result, and collection of the reported outcome []. A examine was categorized as excessive threat if at the very least one area was rated “high risk.” If there have been no high-risk domains however at the very least one area raised considerations, the examine was rated as “some concerns.” Only research with low threat in all domains had been designated as “low risk.”

Effect Measures

The main outcomes (anthropometric measures) and secondary outcomes (dietary measures) had been steady variables. To account for the range in measurement scales throughout research, standardized imply variations (SMDs) with 95% CIs had been used because the abstract impact measure. SMD estimates of 0.20, 0.50, and 0.80 had been interpreted as small, average, and huge impact sizes, respectively []. For outcomes the place a decrease worth signifies scientific enchancment (eg, physique weight, BMI, and waist circumference), the route of the impact dimension was aligned throughout knowledge coding within the Comprehensive Meta-Analysis software program (Biostat) in order that optimistic SMDs constantly signify a larger discount within the intervention group. To keep away from double-counting within the meta-analysis and make sure that every examine contributed unbiased individuals, we solely included one pair of baseline and postintervention knowledge per examine within the main evaluation, chosen in response to the aforementioned hierarchy. SMDs had been calculated utilizing the change from baseline to postintervention for each the intervention and management teams. In circumstances the place a examine reported outcomes for a number of intervention arms in comparison with a single management group, we mixed the intervention teams or cut up the management group to take care of the independence of observations. When change scores weren’t straight reported, they had been calculated utilizing baseline and postintervention means and SDs, assuming a correlation coefficient of 0.5 the place crucial.

Synthesis Methods

Quantitative synthesis was carried out utilizing a random-effects mannequin, as scientific and methodological heterogeneity was anticipated throughout the included stand-alone digital interventions. To enhance the precision of the pooled estimates and reduce the danger of false positives, the HKSJ methodology was utilized.

Statistical heterogeneity was assessed utilizing the Cochran Q statistic and the I² index, with I²>50% indicating substantial heterogeneity []. To additional quantify the dispersion of results in real-world settings, 95% PIs had been calculated alongside the pooled estimates. Subgroup analyses had been prespecified primarily based on the kind of management group (lively vs waitlist), intervention period (≤12 weeks vs >12 weeks), supply sort (app vs web site), and goal life-style (weight-reduction plan vs bodily exercise vs mixed). All statistical analyses had been performed utilizing Comprehensive Meta-Analysis (model 3) software program (Biostat) and R (model 4.5.1; R Foundation for Statistical Computing) with the “meta” package deal.

Reporting Bias Assessment

To assess potential reporting biases, together with publication bias and small-study results, we used each visible and statistical strategies. Funnel plot asymmetry was visually inspected. Additionally, the Egger regression take a look at was carried out to statistically consider small-study results, with a P<.10 indicating important asymmetry []. Furthermore, a trim-and-fill sensitivity evaluation was performed to estimate the potential affect of lacking research on the pooled impact dimension []. We additionally thought of whether or not the noticed asymmetry might be attributed to components aside from publication bias, akin to variations in examine high quality or substantial interstudy heterogeneity.

Certainty Assessment

The Grading of Recommendations, Assessment, Development, and Evaluation framework was used to guage the knowledge of proof throughout 5 domains: threat of bias, imprecision, inconsistency, indirectness, and publication bias. Depending on the standard evaluation inside these domains, the knowledge of proof might be both downgraded or upgraded. The closing classification for every final result was assigned to 1 of 4 ranges: “very low,” “low,” “moderate,” or “high” []. Two unbiased reviewers (SAL and JHP) assessed the proof throughout 5 domains. Any discrepancies within the grading had been resolved via consensus between the two reviewers.

Ethical Considerations

This examine is a scientific evaluate and meta-analysis of beforehand printed literature and didn’t contain direct interplay with human individuals or the gathering of identifiable non-public info. According to institutional and nationwide tips (eg, the Bioethics and Safety Act of the Republic of Korea), the sort of analysis is exempt from Institutional Review Board evaluate because it makes use of publicly obtainable knowledge and carries no threat to people.

Results

Study Selection

A complete of 34,304 literature data had been recognized from the databases, and 12 further data had been recognized via different sources. After eradicating 17,327 duplicate data, 16,977 data had been screened primarily based on titles and abstracts, of which 16,325 had been excluded. The main causes for exclusion at this stage had been (1) clearly irrelevant matters (eg, nondigital scientific therapies or pharmacological interventions), (2) ineligible examine varieties (eg, examine protocols, evaluate papers, or editorials), and (3) ineligible populations (eg, pediatric or adolescent samples) identifiable straight from titles or abstracts. Subsequently, 652 full-text papers had been assessed for eligibility. Of these, 633 papers had been excluded for the next causes: ineligible mixed interventions (n=335), ineligible management teams together with digital interventions (n=229), lack of weight-related final result measures (n=68), and ineligible individuals (n=1). Finally, 19 papers had been included on this systematic evaluate and meta-analysis ().

‎

Study Characteristics

gives traits of the 19 included research. All research had been printed between 2007 and 2024. A complete of 3556 individuals (71% feminine) had been included within the 19 trials, and the entire pattern sizes per examine ranged from 31 to 650 individuals. The imply participant age ranged from 29.4 to 51.2 years. All individuals had been labeled as having obese or weight problems (BMI≥23 kg/m²), with various inclusion thresholds throughout research.

The interventions focused both weight-reduction plan alone or a mix of weight-reduction plan and bodily exercise [-]; notably, no examine focused bodily exercise alone. Common intervention methods included self-monitoring, customized objective setting, and tailor-made suggestions. Delivery platforms included cellular apps (n=7) [,,,,,,], web sites (n=11) [,-,-,,,,], or each (n=1) []. Comparison teams included lively controls (eg, in-person dietary counseling and meals diaries; n=7) [,,,,,,], nonspecific lively controls (eg, schooling or regular care; n=5) [,,,,], and waitlist controls (n=9) [-,,,-]. The period of interventions various from 12 weeks to 24 months, with 12-week applications being the commonest (n=9; ) [,,,,-,,,]. To make sure the stand-alone nature of the included interventions, we meticulously verified the supply mechanisms of all digital interactions. While some research used phrases akin to teaching, suggestions, or peer help, these had been confirmed to be delivered via automated, system-generated algorithms or asynchronous, preprogrammed modules with out real-time human scientific labor ().

Body weight and BMI had been probably the most often reported as main outcomes. Eight research additionally assessed dietary outcomes akin to caloric consumption and macronutrient composition [,-,,,]. Retention charges, as reported in , ranged from 44.1% to 95%, with most research attaining charges above 70%. Mobile app interventions tended to point out barely larger retention in comparison with web-only platforms, though variation existed ().

Risk of Bias in Studies

Among the 19 research, 3 research had been labeled to have a excessive threat of bias, primarily as a result of points in lacking final result knowledge and collection of the reported outcomes. The remaining 16 research had been judged as both low threat (n=7) or elevating some considerations (n=9), with the commonest considerations arising from deviations from meant interventions or ignorance on prespecified evaluation plans ( [-]).

‎

Results of Individual Studies

For all 19 included research (N=3556), the person impact sizes for anthropometric outcomes ranged from SMD −0.04 to 0.72. Similarly, for the 8 research reporting dietary outcomes (n=1365), particular person SMDs confirmed a variety of −0.10 to 0.62. The abstract knowledge for every examine, together with means, SDs, and pattern sizes for each intervention and management teams at baseline and postintervention, are detailed in . The forest plots ( [-] and [,-,,,]) present a visible illustration of the person examine results alongside their 95% CIs.

‎

‎

Results of Syntheses

The meta-analysis utilizing a random-effects mannequin with the HKSJ adjustment demonstrated that stand-alone DLSIs had a statistically important impact on enhancing anthropometric outcomes in comparison with controls (SMD 0.26, 95% CI 0.14‐0.38; P<.001; ). To help scientific interpretation, the pooled SMD was translated into kilograms primarily based on a pooled baseline SD of 16.7 kg, derived strictly from the included research that reported physique weight in kilograms. This corresponds to an extra weight lack of roughly 4.34 kg. Considering the variability in baseline SDs throughout the included research (starting from 10.1 kg to 25.2 kg), the estimated weight reduction might plausibly vary from 2.62 kg to six.55 kg, relying on the inhabitants’s baseline variance. While the I² statistic (56.1%; P<.001) indicated substantial heterogeneity (Q=40.99; P<.001; τ=0.189; τ²=0.036), the 95% PI was calculated as −0.16 to 0.68. This signifies that whereas the typical impact is critical, the result in particular future settings is anticipated to fluctuate, and there’s a risk that the intervention could present no profit and even be much less efficient than the management in sure contexts. For dietary outcomes, stand-alone DLSIs additionally confirmed a major impact (SMD 0.26, 95% CI 0.04‐0.48; P=.008; ). Substantial heterogeneity (I²=57.5%; P=.01) was noticed (Q=16.48; P=.01; τ=0.209; τ²=0.044), and the 95% PI ranged from −0.29 to 0.81, indicating that whereas the typical impact is optimistic, the affect of those interventions in a brand new particular setting might fluctuate and should not all the time obtain a optimistic outcome.

Subgroup analyses had been performed to discover potential moderators of intervention results. First, evaluation by goal habits revealed that interventions specializing in each weight-reduction plan and bodily exercise demonstrated a small important impact on anthropometric outcomes (SMD 0.27, 95% CI 0.12‐0.42; 95% PI −0.04 to 0.58; P<.001; I²=67.5%), whereas diet-only interventions didn’t attain statistical significance (SMD 0.14, 95% CI −0.05 to 0.32; 95% PI −0.07 to 0.35; P=.12; I²=0%). For dietary outcomes, combined-target interventions yielded a major impact (SMD 0.30, 95% CI 0.08‐0.53; 95% PI −0.16 to 0.76; P=.008; I²=62.6%; ).

Second, intervention period was examined. Stand-alone DLSIs ≤12 weeks confirmed a small, important impact on anthropometric outcomes (SMD 0.23, 95% CI 0.07‐0.40; 95% PI −0.17 to 0.65; P=.02; I²=56.1%), corresponding to these >12 weeks (SMD 0.28, 95% CI 0.06‐0.50; 95% PI −0.27 to 0.84; P=.01; I²=60.4%). The take a look at for subgroup variations confirmed no important disparity between short- and long-term durations (P=.71). For dietary outcomes, longer interventions didn’t present a major impact (SMD 0.47, 95% CI −0.21 to 1.16; 95% PI −0.70 to 1.65; P>.05; I²=42.6%), though no important heterogeneity was noticed. Similarly, shorter ones didn’t attain significance (SMD 0.12, 95% CI −0.06‐0.30; 95% PI −0.30 to 0.54; P>.05; I²=0%), regardless of having no heterogeneity (). The formal take a look at for subgroup variations indicated that the distinction between these 2 period classes was not statistically important (P=.44). This means that intervention period doesn’t considerably average the consequences on dietary outcomes, and stand-alone DLSIs didn’t yield a constant important affect on weight-reduction plan whatever the program size.

Third, by management sort, stand-alone DLSIs confirmed no important results vs lively controls on anthropometric outcomes (SMD 0.04, 95% CI −0.08 to 0.16; 95% PI −0.08 to 0.16; P=.52; I²=0.0%) and weight-reduction plan (SMD 0.23, 95% CI −0.23‐0.70; 95% PI −0.46 to 0.92; P=.33; I²=64.4%). Compared to nonspecific lively controls, the impact on anthropometric outcomes was small however important (SMD 0.13, 95% CI 0.00‐0.26; 95% PI 0.00 to 0.26; P=.05; I²=0%). The strongest results had been noticed vs waitlist controls, with a average impact on anthropometric outcomes (SMD 0.57, 95% CI 0.42‐0.73; 95% PI 0.42 to 0.72; P<.001; I²=0%) and a major impact on weight-reduction plan (SMD 0.36, 95% CI 0.11‐0.62; 95% PI 0.11 to 0.61; P=.006; I²=44.2%). A major subgroup distinction was noticed for anthropometric outcomes (P<.001), indicating management sort as a major moderator ().

Fourth, by supply sort, interventions delivered by way of cellular apps confirmed no important impact on anthropometric outcomes (SMD 0.24, 95% CI 0.00‐0.49; 95% PI −0.20 to 0.68; P=.05; I²=62.3%), whereas a major impact was noticed for dietary outcomes (SMD 0.33, 95% CI 0.02‐0.64; 95% PI −0.21 to 0.87; P=.03; I²=73.6%). Website-based interventions additionally demonstrated a major impact on anthropometric outcomes (SMD 0.23, 95% CI 0.08‐0.37; 95% PI −0.03 to 0.49; P=.004; I²=60.2%), however didn’t attain statistical significance for dietary outcomes (SMD 0.22, 95% CI −0.04‐0.48; 95% PI −0.04 to 0.48; P=.10; I²=22.7%; ). No important subgroup variations had been discovered between app and web site supply for both anthropometric (P=.90) or dietary outcomes (P=.61; ).

Sensitivity analyses had been performed by aggregating a number of impact sizes inside the similar examine utilizing their arithmetic common to guage the robustness of the first findings in opposition to potential unit-of-analysis errors. The important results of stand-alone DLSIs on each anthropometric and dietary outcomes remained sturdy. For anthropometric outcomes, the pooled impact dimension barely elevated (SMD 0.27, 95% CI 0.14‐0.41; P<.001; 95% PI −0.15 to 0.70; I²=56.7%), whereas for dietary outcomes, the impact dimension decreased modestly (SMD 0.22, 95% CI 0.02‐0.42; P=.03; 95% PI −0.32 to 0.75; I²=54.7%; ). In addition, a secondary sensitivity evaluation was carried out by excluding 3 research recognized as having a excessive threat of bias whereas sustaining the study-level aggregation of impact sizes. For anthropometric outcomes, the pooled impact dimension remained statistically important (SMD 0.27, 95% CI 0.14‐0.39; P<.001; 95% PI −0.21 to 0.74; I²=88.1%). Similarly, for dietary outcomes, the impact remained important even after excluding lower-quality proof (SMD 0.20, 95% CI 0.01‐0.39; P=.04; 95% PI −0.37 to 0.77; I²=64.2%).

Furthermore, to make sure the findings weren’t skewed by the inclusion of long-term trials, a 3rd sensitivity evaluation was restricted to trials with a period of ≤6 months. For anthropometric outcomes, the impact remained extremely constant (SMD 0.26, 95% CI 0.13‐0.39; P<.001; 95% PI −0.22 to 0.74; I²=88.1%). Similarly, for dietary outcomes, the numerous profit was maintained (SMD 0.26, 95% CI 0.04‐0.48; P=.03; 95% PI −0.34 to 0.86; I²=64.3%; ). These cumulative outcomes point out that the general conclusions weren’t delicate to the presence of a number of outcomes per examine, the inclusion of research with a excessive threat of bias, or variations in trial period, additional confirming the steadiness and reliability of the principle evaluation.

Reporting Biases

Small-study results had been assessed for important ends in the principle evaluation utilizing funnel plots and the Egger take a look at. For anthropometric outcomes, the Egger take a look at yielded an intercept of 1.35 (95% CI –1.54 to 4.24; t₁₇=0.946; P=.35), and for dietary outcomes, the intercept was 0.28 (95% CI –1.19 to 1.75; t₆=0.108; P=.92), indicating no statistical proof of asymmetry in both case ( and ). However, particularly for anthropometric outcomes, visible inspection prompt potential asymmetry within the lower-left quadrant, presumably as a result of heavy reliance on waitlist-controlled trials and the danger of unpublished small-scale trials. To tackle this, a trim-and-fill sensitivity evaluation was carried out (), which recognized 2 probably lacking research on the left facet of the funnel. After imputing these research, the adjusted pooled impact dimension remained statistically important (adjusted SMD 0.21, 95% CI: 0.09‐0.34; P<.001). The heterogeneity for the adjusted mannequin was noticed (Q₂₀=120.54; P<.001; τ=0.294; τ²=0.086), confirming the robustness of the first findings. Consequently, the noticed impact sizes for each outcomes are unlikely to be pushed primarily by publication bias.

‎

‎

Certainty of Evidence

Based on the Grading of Recommendations, Assessment, Development, and Evaluation evaluation of every important impact estimate, this examine concludes that stand-alone DLSIs present a moderate-certainty profit for each anthropometric and dietary outcomes. The certainty of proof for anthropometric outcomes was downgraded by one degree as a result of severe considerations concerning inconsistency. This determination was made due to the substantial statistical heterogeneity. Furthermore, our subgroup evaluation revealed that the kind of management group was a key supply of heterogeneity, indicating that the noticed dispersion was largely defined by variations in examine design reasonably than inconsistent intervention results. Similarly, the knowledge of proof for dietary outcomes was downgraded by one degree as a result of severe inconsistency amongst research, regardless of a constant route of results ().

Meta-Regression

Meta-regression evaluation confirmed that retention charges affect each anthropometric (β=0.005, 95% CI 0.0001‐0.0100; P=.05) and dietary outcomes (β=0.0175, 95% CI 0.0103‐0.0246; P<.001), indicating that larger participant retention was related to larger intervention effectiveness ( and ). This means that a rise in participant retention was related to a corresponding improve within the SMD, highlighting the significance of methods to reduce dropout in digital interventions.

‎

‎

Discussion

Interpretation

This systematic evaluate and meta-analysis evaluated the efficacy of stand-alone DLSIs on anthropometric and dietary outcomes amongst adults with obese or weight problems, with a rigorous concentrate on isolating the unbiased therapeutic potential of digital know-how. Our findings show that stand-alone DLSIs can obtain statistically important enhancements in each anthropometric and dietary outcomes, supporting their function as unbiased interventions. Notably, the noticed magnitude of weight reduction exceeds the standard threshold for a minimal clinically vital distinction in weight problems administration, which is usually cited as 2‐3 kg []. This means that stand-alone digital platforms can obtain clinically significant outcomes corresponding to extra resource-intensive conventional applications, reinforcing their viability as a scalable public well being device, particularly in resource-constrained settings [].

To our data, this examine is progressive as it’s the first to strictly delineate the impact of DLSIs by eliminating adjunctive human interplay. Distinct from earlier meta-analyses that usually included multimodal or hybrid interventions [,], this examine clarifies the particular therapeutic potential of digital know-how alone. Prior analysis incorporating structured group codecs or peer interactions typically reported numerically bigger impact sizes [,]. However, our outcomes verify that stand-alone DLSIs nonetheless yield statistically important advantages, suggesting that the digital part itself is a sturdy driver of change. This reinforces the utility of those interventions in environments the place scientific labor or real-time moderation is unavailable [].

To guarantee a balanced interpretation of those outcomes, a number of analytical components have to be thought of. First, the noticed statistical heterogeneity displays the inherent variety in digital well being implementation []. Notably, our subgroup analyses clarified that this variance was primarily pushed by comparator depth, as heterogeneity was totally resolved within the active-control subgroup. Interventions focusing on each weight-reduction plan and bodily exercise demonstrated synergistic results, probably as a result of extra complete behavioral self-regulation [,]. Regarding intervention period, an attention-grabbing discrepancy emerged: stand-alone DLSIs demonstrated constant and important advantages for anthropometric outcomes no matter program size, indicating a sturdy affect on bodily parameters. In distinction, dietary outcomes didn’t present statistically important enhancements in both short-term or long-term subgroups. The lack of serious impact throughout all timeframes means that stand-alone DLSIs could face challenges in independently facilitating significant dietary adjustments, whatever the period supplied [,]. This discrepancy between anthropometric success and dietary stagnation could suggest that whereas digital instruments successfully promote weight reduction—probably via enhanced self-monitoring or bodily exercise monitoring—they could require further built-in help, akin to environmental modifications, to beat the complexities of dietary behavior transformation []. Furthermore, effectiveness adopted a gradient primarily based on comparator sort, being simplest in opposition to waitlist controls and diminishing in opposition to structured lively comparators, probably reflecting nonspecific intervention results akin to consideration or construction [,]. Meta-regression analyses additional revealed that larger retention charges had been considerably related to larger enhancements in each outcomes, underscoring that sustained consumer engagement is a crucial moderator of effectiveness [,].

Furthermore, the danger of bias inside the proof base have to be famous, as roughly 63.2% of the included research had been rated as having excessive threat or some considerations. Sensitivity analyses, nevertheless, supported the robustness of those findings, as excluding research with a excessive threat of bias preserved the importance of the principle outcomes whereas resulting in a marked discount in heterogeneity, thereby reinforcing the reliability of the noticed results throughout higher-quality trials.

The credibility of those outcomes is additional mirrored within the certainty of proof. The certainty for each anthropometric and dietary outcomes was rated average and was primarily downgraded as a result of severe inconsistency. This average certainty signifies that whereas the present proof is promising, additional high-quality analysis might be priceless to extend the precision of those estimates and make sure the steadiness of the noticed results throughout numerous intervention methods.

Importantly, the interpretation of those findings requires a transparent distinction between the typical impact and the distribution of results throughout totally different settings. While the 95% CIs verify a major common impact, the calculated 95% PIs for each outcomes cross the null line. This crucial distinction means that whereas stand-alone DLSIs are efficient on common, the anticipated impact in a selected future scientific setting could also be much less sure, with the potential for null and even adverse outcomes relying on the implementation context [].

This examine affords a number of methodological and sensible strengths. First, it’s the first meta-analysis to isolate the consequences of DLSIs, thereby eliminating potential confounding from adjunctive intervention codecs. Second, by evaluating each anthropometric and dietary outcomes throughout numerous supply modes and comparator varieties, this examine gives a complete and scalable understanding of stand-alone DLSI effectiveness []. Third, the inclusion of detailed subgroup and meta-regression analyses helped make clear the circumstances below which stand-alone DLSIs are simplest—notably in relation to focused behaviors, intervention period, and comparator group. Fourth, the unique inclusion of RCTs, mixed with rigorous high quality assessments and sensitivity analyses, enhances the methodological credibility of the findings. Finally, the average certainty of proof for each anthropometric and dietary outcomes additional underscores the reliability of the noticed results, reinforcing the potential of stand-alone DLSIs as evidence-based instruments for weight problems administration [].

Limitations

Nevertheless, a number of limitations ought to be acknowledged when decoding these findings. First, most included trials featured brief intervention intervals (sometimes ≤12 weeks), which limit conclusions concerning the long-term upkeep of weight-related adjustments. Second, substantial heterogeneity was not constantly defined by system sort or behavioral change strategies, suggesting that the variance could stem from unmeasured variables akin to individual-level consumer interplay patterns. Third, the dearth of standardized reporting on intervention depth and cost-effectiveness hinders direct translation into scalable well being coverage. Fourth, sustaining consumer engagement stays an inherent problem in unsupervised digital applications []. Fifth, adherence and consumer engagement—crucial determinants of digital intervention success—had been underreported or inconsistently measured throughout trials, precluding formal evaluation of those as moderators. Sixth, the presence of potential publication bias for sure outcomes means that the noticed impact sizes could be barely overestimated, necessitating a cautious interpretation []. Finally, the absence of formal cost-effectiveness evaluations inside the included research limits the interpretation of findings into scalable well being coverage suggestions.

To tackle these limitations, future analysis ought to prioritize long-term RCTs that may consider the sustainability of DLSI-induced weight adjustments past the standard short-term window. Additionally, extra granular analyses are wanted to isolate the lively parts and supply mechanisms—akin to platform sort or particular behavioral change strategies—that contribute most importantly to intervention efficacy. Special consideration ought to be given to creating and testing customized options, akin to AI-driven tailoring that adapts to real-time consumer knowledge, and the cultural adaptation of content material to fulfill the particular wants of numerous ethnic and linguistic populations []. Given the underreporting of adherence and engagement metrics in present trials, future research ought to combine standardized measures of consumer interplay and take a look at the affect of adaptive methods, together with AI-driven personalization, on sustaining consumer retention and enhancing outcomes. Finally, incorporating cost-effectiveness analyses will present crucial perception for policymakers contemplating the adoption of stand-alone DLSIs in routine care and public well being programming.

Implications

This systematic evaluate and meta-analysis gives a number of crucial implications for the sector of digital well being and weight problems administration. First, by way of innovation and distinctiveness, our examine is exclusive in that, in contrast to earlier opinions that usually conflated purely digital instruments with hybrid fashions involving in-person teaching, it remoted the stand-alone impact of DLSIs with out individualized scientific help. This methodological focus distinguishes our evaluate from present opinions by strictly excluding human-contact confounding components to show that digital know-how alone can obtain statistically important weight reduction and dietary enhancements [,]. Second, concerning its contribution to the sector, our findings spotlight that whereas the general impact dimension of stand-alone interventions could also be smaller than that of human-supported ones, they provide a extremely scalable and cost-effective answer for public well being. This examine particularly identifies that stand-alone DLSIs are simplest when in comparison with minimal-intervention or waitlist controls, offering a transparent benchmark for future digital well being device growth. Finally, for real-world implications, these outcomes recommend that stand-alone DLSIs can function an accessible first-step intervention in a stepped-care mannequin for weight problems []. For clinicians and policymakers, which means that stand-alone digital platforms could be broadly deployed to populations with restricted entry to intensive in-person counseling, providing a viable technique to scale back the worldwide burden of obesity-related ailments [,].

Conclusions

Our examine validates stand-alone DLSIs as efficient main instruments for weight problems administration, providing moderate-certainty proof free from direct human-contact confounding. This evaluate contributes to the sector by validating technology-driven interventions as efficient main instruments. In the true world, these outcomes suggest that stand-alone DLSIs provide a scalable first step in stepped-care fashions, offering an accessible technique for clinicians and policymakers to handle the worldwide weight problems burden. However, given the substantial heterogeneity and 95% PIs crossing the null, these common results shouldn’t be considered as assured advantages. The context-dependent efficacy of stand-alone DLSIs means that whereas they’re a viable public well being various, their implementation requires cautious monitoring for consistency throughout totally different settings.