Abstract
Recent advancements in smartphone and tablet-based 3D and 4D scanning technologies offer a promising alternative to traditional anthropometric methods, such as dual-energy X-ray absorptiometry (DXA), by enabling the generation of digital human models using accessible and cost-effective devices. This systematic review aims to evaluate the accuracy, validity, and reliability of anthropometric measurements obtained through smartphone and tablet-based 3D and 4D scanning technologies compared to conventional methods. A total of 2172 studies, no Meta-analysis, systematic reviews or reviews, were screened on March 11, 2025, across five databases (PubMed, Web of Science, Medline, Cochrane, and ScienceDirect), with 13 studies meeting the inclusion criteria. Included studies focused on healthy people above or 18 years old with only smartphone and tablet-based measurements and nothing like facial, dental dimensions or invasive techniques such as MRI and CT. Methodological quality was assessed by using the Joanna Briggs Institute appraisal tools. Key outcomes extracted include fat mass, body fat, body mass, fat free mass, waist-hip-ratio, appendicular lean mass and body height. Overall, reliability, for key anthropometric parameters such as fat mass, were strong (ICC > 0.9), validity high (R2 ≥ 0.837), and good accuracy (RMSE 2.5–5.2). Although minor inconsistencies were noted in some fat-related metrics. Although this review aims to assess both 3D and 4D components, no studies using 4D measurements were found. To conclude, smartphone and tablet-based 3D scanning demonstrates strong potential as a viable and cost-effective alternative to traditional anthropometric methods. However, recommendations must be considered such as further algorithm development and establishment of standardized protocols.
Keywords: 3D scanners, Smartphones, Anthropometry, Measurements, Size and shape
Introduction
The field of anthropometry, dedicated to the measurement of human body size and shape, is a critical tool for assessing body composition and predicting health outcomes [1]. Anthropometric data provide indirect but valuable indicators of internal composition, such as fat mass, lean tissue, and overall body fat percentage [1, 6]. These metrics are increasingly important in public health for identifying at-risk individuals, monitoring nutritional status, and evaluating interventions in obesity and metabolic disorders [2]. In sports science, anthropometric parameters are used to link body composition with athletic performance, for example in endurance cycling [3]. The advent of advanced 3D body scanning technologies has further increased precision, enabling accurate morphological analyses and estimations of energy expenditure during exercise [4]. Beyond diagnostics and performance monitoring, 3D scanning is also applied in design and production, such as creating personalized equipment through 3D printing (e.g., custom-fitted facepieces) [5]. Finally, the integration of mobile health technologies introduces opportunities to improve access to care, particularly for individuals limited by transportation or geographical barriers [6].
In this review, traditional anthropometric methods refer to techniques that have been widely used for decades to assess body size, proportions, and composition before the advent of 3D and smartphone imaging. These include direct manual methods (e.g., tape measures, calipers, rulers) and indirect methods (e.g., bioelectrical impedance analysis and the four-compartment model). The 4C model is considered the most accurate indirect method, as it integrates body density, total body water, bone mineral content, and residual mass to reduce error compared to single-technique estimates [8]. Manual techniques remain common due to their simplicity and low cost [7, 9]. However, they are examiner-dependent, time-consuming, and require strict calibration, consistent application, and correct identification of anatomical reference points to ensure reliability [7]. Moreover, they are typically limited to one-dimensional data such as height, mass, and circumference, and when extended to 3D models, they rely on estimations that lack the fine detail needed for precise analysis [10, 11]. Indirect methods, while more accurate, often require multiple measurements and are impractical for routine or large-scale use. Radiographic imaging and two-dimensional photography have also been applied in the past, but both approaches are limited in their ability to represent true body morphology [9, 12]. While 3D surface scanners can generate highly accurate avatars [13, 14], they remain expensive and non-portable, restricting their accessibility. Similarly, dual-energy X-ray absorptiometry (DXA) is still regarded as the clinical gold standard for body composition analysis due to its precision and speed, but its high cost, size, and lack of portability make it impractical for daily or large-scale us [21]. Although advanced scanning can capture changes in body shape over time, these costs remain a major barrier. Smartphone and tablet-based imaging have therefore been proposed as a potential bridge to more accessible 3D analysis, with built-in cameras enabling dynamic body movement recording [16].
Digital human models (DHM) are increasingly integrated into smartphone and tablet-based 3D imaging systems to create accurate full-body avatars, enabling precise and non-invasive assessment of anthropometric and body composition parameters [21]. Smartphone and tablet-based 3D imaging has emerged as a promising alternative to traditional methods [17]. These systems offer detailed depictions of external body morphology while avoiding the high costs and lack of portability of approved 3D scanners [14]. They allow fast, mobile, and low-cost data collection [15], and advances in smartphone and tablet technology now enable 3D scans within seconds and at a fraction of the cost of traditional systems [18]. In addition to affordability and speed, these methods are efficient in data acquisition and processing, easy to use, and applicable across a wide range of settings [19]. AI-powered applications further improve accuracy, automation, and adaptability in both clinical and research environments [20].
Traditional reference methods such as DXA and the 4C model are widely used to assess body composition due to their high precision, but their cost and lack of portability make them impractical for routine use. Smartphone and tablet-based 3D scanners offer a promising, low-cost, and accessible alternative with potential for both clinical and non-clinical applications. Despite this, there is currently no clear consensus on their accuracy, reliability, or validity. The wide variety of available applications also highlights the need for further technological innovation to improve accuracy and precision [20]. While multiple studies have shown that smartphone and tablet-based methods can reliably create 3D humanoid avatars [21], validation efforts remain limited, focusing mostly on static models and rarely addressing dynamic or real-life conditions [22]. This gap underscores the need for systematic evaluation of the current evidence on the accuracy, validity, and reliability of smartphone and tablet-based anthropometric methods.
This systematic review aims to systematically evaluate the reliability, accuracy, and validity of smartphone and tablet-based full-body imaging compared with traditional anthropometric techniques. In addition, it considers the potential of smartphone and tablet-based systems to provide quantitative measurements such as body mass, body fat percentage, and fat-free mass. To ensure objectivity and minimize bias, the evaluation is structured within a clear psychometric framework. Reliability is defined as the ability of smartphone and tablet-based imaging methods to produce consistent results. It depends both on the technology itself and on procedural consistency, including differences over time, between observers, or due to variations in subject posture during scanning [23]. The type of reliability assessment (intra-rater, within a single session, or inter-rater, across sessions or observers) is considered to provide a comprehensive evaluation [24]. Accuracy refers to the absolute agreement between measurements obtained with smartphone and tablet-based 3D/4D imaging and those from traditional methods such as DXA or manual tape assessments. In terms of concurrent validity, this review examines the extent to which smartphone and tablet-based imaging correlate with traditional anthropometric techniques.
Methods
This systematic review was registered in PROSPERO (ID: CRD420251009296) and conducted on March 11, 2025, following PRISMA guidelines [25].
Eligibility criteria
Search results were fist imported into Endnote [26] to remove duplicates and subsequently screened in Rayyan [27]. Inclusion and exclusion criteria were defined using the PICOST framework (Table 1). Eligible studies included smartphone and tablet-based 3D imaging and reported whole-body anthropometric outcomes. Due to rapid changes in body measurements of children, only individuals that are older or 18 years were assed to minimize potential bias.
Table 1.
Eligibility criteria following the PICO(-ST) method
| Inclusion | Exclusion | |
|---|---|---|
| P | Human adults (> 18 years) | Human (< 18 years); animals |
| I |
smartphone, tablet Dual-energy X-ray absorptiometry (DXA) |
X-ray without DXA MRI; CT |
| C | 3D Scanners and tape measure | Surgery |
| O | Whole body; shape; body shape, digital human body | Dental; geographical mapping; non-digital human; body parts; facial; skeleton body |
| S | Randomized Controlled Trial; cohort study; cross-sectional; pilot | Meta-analysis; systematic review; review |
| T | Data range from 2014–2025 | Data before 2014 |
| L |
Full text written in English and Dutch |
Full text written in any other language except English and Dutch |
P: population; I: intervention; C: comparison; O: outcome; S: study design; T: time frame;
L: language
aTANITA model BC-418 MA is a segmental body analysis scale
Reference standards included tape measurements, 3D scanners and DXA. Although DXA is not a 3D scanner, it is widely regarded as the gold standard for assessing body composition and is therefore commonly used as a reference. Excluded were children, animals and invasive methods. Although anthropometric measurements have been applied in fields such as oral and maxillofacial and plastic surgery, smartphone-based surface imaging holds broader potential for rapid assessment of body composition and morphology. Therefore, these were excluded.
Only full-text articles in English or Dutch, published between 2014 and 2025, were included. This period marks the advent and clinical research integration of commercially available smartphone-based 3D imaging technologies. No articles were excluded by language.
Data availability
Five databases were searched: PubMed [28] (Table 2), Web of Science [29] (Table 3), Medline [30] (Table 4), Cochrane [31] (Table 5), and ScienceDirect [32] (Table 6). In PubMed, MeSH terms were combined with free keywords, using truncation to broaden the search. The strategy was first developed for PubMed (Table 2) and adapted for other databases. The final search was conducted on March11, 2025.
Table 2.
Database Search Strategy in PubMed
| Database: PubMed | Search strategy |
|---|---|
| P | / |
| I | ((smartphone app) OR (iPhone app) OR mobile OR (smartphone application*) OR (iPhone application*) OR (smartphone-based imaging) OR ("Mobile Applications"[Mesh]) OR (mobile applications)) |
| C | (scanning OR imaging OR photogrammetry OR infrared OR (Depth camera)) |
| O | ((anthro*) OR (metric*) OR (meas*)) AND ((Concurrent Validity) OR Validity OR reliability OR accuracy OR (landmark position error) OR (Reproducibility of results) OR comparison OR (clustering high-dimensional data) OR ("Reproducibility of Results"[Mesh]) OR (reproducibility of results)) AND (3D OR ("Imaging, Three-Dimensional"[Mesh]) OR (imaging, three-dimensional) OR 4D OR (digital human modeling) OR dynamic OR DHM OR (body shape) OR (geometric*)) |
P: population; I: intervention; C: comparison; O: outcome
Table 3.
Database Search Strategy in Web of Science
| Database: web of science | Search strategy |
|---|---|
| P | / |
| I | ALL = ((smartphone app) OR (iPhone app) OR mobile OR (smartphone application*) OR (iPhone application*) OR (smartphone-based imaging)) |
| C | ALL = (scanning OR imaging OR photogrammetry OR infrared OR (Depth camera)) |
| O | ALL = ((anthro*) OR (metric*) OR (meas*)) AND ALL = ((Concurrent Validity) OR Validity OR reliability OR accuracy OR (landmark position error) OR correlation OR (Reproducibility of results) OR comparison OR (clustering high-dimensional data)) AND ALL = (3D OR 4D OR (digital human modeling) OR dynamic OR DHM OR (body shape) OR (geometric*)) |
P: population; I: intervention; C: comparison; O: outcome
Table 4.
Database Search Strategy in Medline
| Database: medline | Search strategy |
|---|---|
| P | / |
| I | ((smartphone app) OR (iPhone app) OR mobile OR (smartphone application*) OR (iPhone application*) OR (smartphone-based imaging) OR (mobile applications)) |
| C | (scanning OR imaging OR photogrammetry OR infrared OR (Depth camera)) |
| O | ((anthro*) OR (metric*) OR (meas*)) AND ((Concurrent Validity) OR Validity OR reliability OR accuracy OR (landmark position error) OR (Reproducibility of results) OR comparison OR (clustering high-dimensional data) OR (reproducibility of results)) AND (3D OR OR (imaging, three-dimensional) OR 4D OR (digital human modeling) OR dynamic OR DHM OR (body shape) OR (geometric*)) |
P: population; I: intervention; C: comparison; O: outcome
Table 5.
Database Search Strategy in Cochrane
| Database: cochrane | Search strategy |
|---|---|
| P | / |
| I | mobile |
| C | imaging |
| O | anthro* AND meas* |
P: population; I: intervention; C: comparison; O: outcome
Table 6.
Database Search Strategy in ScienceDirect
| Database: ScienceDirect | Search strategy |
|---|---|
| P | / |
| I | mobile application |
| C | imaging |
| O | anthropometric AND measurement AND results AND 3D |
P: population; I: intervention; C: comparison; O: outcome
Search strategy
The search strategy covered four domains: (1) 3D/4D scanners, (2) smartphones, (3) anthropometry, and (4) body size/shape. Relevant MeSH terms and free-text synonyms were combined using Boolean operators (‘AND’, ‘OR’) to maximize sensitivity (e.g., “digital human models,” “smartphone-based imaging,” “body morphology”). This strategy was applied across all databases with minor adaptations, using a publication date filter (2014–2024).
Selection process
A two-step double-blind approach was used. In phase one, three reviewers (E.V.D.H., M.S., F.V.O.) independently screened titles and abstracts, resolving disagreements through discussion or a third reviewer. In phase two, full texts of potentially eligible studies were similarly assessed, with final decisions confirmed by senior reviewers (S.S. and S.T.).
Data collection process
Study selection and data collection followed PRISMA guidelines and are shown in Fig. 1. A total of 2172 records were retrieved (PubMed: 438; Web of Science: 1235; Medline: 157; Cochrane: 46; ScienceDirect: 296). After removing 332 duplicates, thirteen articles [21, 33–44] were included following full-text review.
Fig. 1.
Flowchart 2000
Data items
Extracted data included study design, population, measurement methods, anthropometrics, and metrics for concurrent validity, reliability, and accuracy. Core outcomes were fat mass, body fat, body mass, fat-free mass, waist-to-hip ratio, appendicular lean mass, and body height. Sixteen additional outcomes appeared in only one study, including total lean mass, visceral fat, body volume, and joint angles. Reliability was assessed via ICC, CCC, and 95% CI; validity via r, R2, and p; and accuracy via RMSE, RMSE-CV, LOA, and MAE. Table 7 presents the evidence, while Table 8 provides operational details for reliability, validity, and accuracy.
Table 7.
Table of evidence for reliability, concurrent validity and accuracy
| Study, year | Study design | Population n; age (year ± SD); gender | Measuring methods | Anthropometrics | Concurrent validity & reliability | Accuracy |
|---|---|---|---|---|---|---|
| Farina et al. [33] | Cross-sectional |
n = 117; (63F, 54 M) (38.7 ± 13.8) FY (32.5 ± 9.8) MY |
CG: DXA EG: DP (Android version 4.2.2, Huawei G730; iOS 9.2, iPhone 5 s) |
Weight; heights; BMI; FFM; FM; BF | R2; ICC; CCC | LOA; RMSE |
| Smith et al. [34] | Cross-sectional |
n = 59; gender: NA (40.0 ± 0) SD; NA |
CG1: SS20 scanner CG2: tape measurements EG: MTS (iPhone X) |
Girth: Waist; hip; Thigh (L-R); Arm (L-R); body height |
p; R2; ICC | LOA; RMSE |
| Florez et al. [35] | Cross-sectional |
n = 96; (51F-45 M) (23.7 ± 6.5) Y SD; NA |
CG1: 4C CG2: DXA EG: Army Men1 (iPhone 13 Pro Max) |
BF%; FM; FFM | r; CCC | LOA; MAE; RMSE |
| Bušić et al. [36] | Cross-sectional |
n = 51; (12F-39 M) Age; NA SD; NA |
CG: classical anthropometry EG: portable 3D scanner (iPad Air 2) |
Body height; chest girth; breast girth; hips girth; waist girth; L-R upper arm girth; L-R elbow girth.; L-R forearm girth; L-R wrist girth; L-R upper leg girth; L-R lower leg girth | p; R2; r | RMSE |
| Choudhary et al. [37] | Cross-sectional |
n = 550; (280F-270 M) Age, NA SD; NA |
CG: self-measured EG: Measure Net (a modified Resnet-18 network) |
WHR | R2 | MAE |
| Smith et al. [38] | Experimental |
n = 38; (21 M-17F) Age; NA SD; NA |
CG: DXA EG: Body scan SPA (iPhone 13) |
BM; BF; FFM; Waist girth | p; CCC; 95%CI | NA |
| Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | Experimental |
n = 16; gender: NA Y = 25 SD; NA |
MSTS (iPad-based 3D mobile scanning tool, Structure SensorTM (2018 version) | LBC; TK; NC; SE; LPT; FKA; SP | ICC; 95% CI | NA |
| Qiao et al. [40] | Observational cohort |
n = Fenland study (CG) 12 435; gender: NA Y = 48–56 SD; NA n EG = 119; (39 M-80F) Y; NA SD; NA |
CG: DXA; TANITA model BC-418 MAa; SECA 2401; measuring tape EG: 3D body-shape app (iPhone X) |
FM; BF; android FM; gynoid FM; visceral FM; abdominal SCAT mass; peripheral FM; TLM; ALM; ALMI | R2 | LOA; RMSE |
| Graybeal J. et al. [41] | Cross-sectional |
n = 123 (54 M-69F) Y = 28 SD; NA |
CG: MFB EG: MTS (IPhone X, IOS v15.6.1) |
BF; BMC VAT; LBM; LBM arm; LBM leg; BSA; BV; SAT; BVT; SAA; BVA; SAL; BVL: head girth; collar girth; chest girth; shoulder girth; UA girth; biceps girth; FA girth; wrist girth; arm girth; waist.; hip.; thigh girth; knee girth; calf girth; ankle girth; arm length: outside leg length | p; R2 | LOA; RMSE |
| McCarthy et al. [42] | Cross-sectional |
n1 (SS20) = 322; (178F-144 M) (47.3 ± 17.6) FY; (45.0 ± 17.0) FM SD; NA n2 (MTS) = 53 (27F-26 M) 38.6 ± 15.3(FY) 39.0 ± 14.1(FM) SD; NA |
CG: SS20 (software version 6.2.1) + DXA EG: MTS (iPhone X) + DXA |
Height; weight; BMI; ALM; head girth; collar girth; chest girth; forearm girth; upper arm girth; waist girth; hip girth; thigh girth; ankle girth; arm length.; outside leg length; SA arm; SA torso; SA leg; SA total; arm V; torso V; leg V; total V | p; R2; 95%CI | RMSE |
| Tinsley M. et al. [21] | Cross-sectional |
n = 131; (58F-73 M) Age; NA SD; NA |
CG: DXA EG1: smartphone 3D scanning app; BF% ADAM EG2: smartphone 3D scanning app; BF% COCO (13 Pro Max, iOS v. 16.5; iPhone 14 Pro, iOS v. 16.6) |
Height; weight; BMI, BF% | p; r; ICC; CCC; 95%CI | RMS; MAE |
| Graybeal J.; Brander F.; Tinsley M. [43] | Cross-sectional |
n = 102; (63F-39 M) Y = 18–75 SD; NA |
CG: 4C model EG1: MTS (iPhone 12 pro) EG2: MTS (Samsung Galaxy S21) |
BF%; FM; FFM | p; r; R2, ICC | LOA; RMSE |
| Graybeal J.; Brander F.; Tinsley M. [44] | Cross-sectional |
n = 184; (114F-70 M) Y = 18–75 SD; NA |
CG = 4C model EG1 = Amazon Halo app iPhone 12 pro, iOS 15.0.1) EG2 = Amazon Halo app (Samsung Galaxy S21, One UI version 3.1, 4.0, and 4.1 and Android version 11 and 12) EG3 = MyBVI app (iPhone 12, iOS 15.0.1) |
BM; TBW; BMC; height; weight; BMI; BF%; FM; FFM | R2; ICC; 95%CI | LOA; RMSE;RMSE-CV |
4C: 4-compartment; 95% CI: 95% confidence interval; ADAM: automatic detection of athlete mode; ALM: appendicular lean mass; ALMI: appendicular lean mass index; app: application; Army Men: development by 3DO (three dimensional optical imaging); BF: body fat; BF%: body fat percentage; BM: body mass (kg); BMC: Bone Mineral Content (cm2); BMI: body mass index; BSA: body surface area (cm2); BV: body volume (cm2); BVA: body volume arms; BVL: body volume legs; BVT: body volume torso; CCC: concordance correlation coefficient; CG: control group; COCO: compound circumference only; DIA: digital imaging analysis; DP: digital image photography; DXA: dual-energy X-ray absorption; EG: experimental group; FA: forearm; FFM: fat free mass (kg); FM: fat mass (kg); FKA: frontal knee angle (°); LBC: external low back curvature (°); LBM: Lean body mass (kg); LOA: limits of agreement; LPT: lateral pelvic tilt (°); MAE: mean absolute error; MFB: mobile fit booty; MSTS: novel three-dimensional imaging mobile surface topography system; MTS: MeThreeSixty, (stationary, clinical device); MyBVI: Body Volume Index (app); NA; not available; NC: external neck curvature (°); P: significant; R2: correlation; RMSE-CV: root mean square error coefficient of variation (%); RMSE: root mean square error; SA: surface area; SAA: surface area arms; SAL: surface area legs; SAT: surface area torso; SE: shoulder elevation (°); SP: scapular prominence (°); SS: size stream; TBW: total body water; TK: thoracic kyphosis (°); TLM: total lean mass; UA: upper arm; VAT: Visceral Adipose Tissue (cm2); V: volume
Table 8.
Table of evidence for reliability, validity and accuracy
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE (%) | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| FM (kg) | Farina et al. [33] |
F; 0.714 M; 0.838 |
F; 0.991 M; 0.982 |
F; 0.952 M; 0.974 |
F; 0.952 M; 0.974 |
F; (− 5.4)–5.6 M; (− 5.7)–5.6 |
F; 3.93 M; 3.12 |
||||
| Florez et al. [35] | 0.97 | 0.94 |
EG-CG1: 0.86 EG-CG2: 0.86 |
(− 5)–6 |
EG-CG1: 3.6 EG-CG2: 3.6 |
||||||
| Qiao et al. [40] | 0.96 | 2.5 | |||||||||
| Graybeal J.; Brander F.;Tinsley M. [43] | CG-EG1: < 0.001 | CG-EG1: 0.85 | CG-EG1: 0.91 | EG1-EG2: | CG-EG1: 7.6 | CG-EG1: 3.86 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Graybeal J.; Brander F.; Tinsley M. [43] | CG-EG2: < 0.001 | CG-EG2: 0.95 | CG-EG2: 0.91 | 0.997–1.000 | CG-EG2: 7.7 | CG-EG2: 3.93 | |||||
| Graybeal J.; Brander F.; Tinsley M. [44] |
CG-EG1: 0.941 CG-EG2: 0.938 CG-EG3: 0.837 |
EG1-EG2: 0.999 EG3: 0.997 |
CG: 21.3–24.8 EG1: 22.2–25.6 EG2: 22.2–25.6 EG3: 22.2–25.6 |
CG-EG1: 5.8 CG-EG2: 5.8 CG-EG3: 10.1 |
CG-EG1: 3.1 CG-EG2: 3.1 CG-EG3: 5.2 |
EG1-EG2: 3.3 EG3: 4.3 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| BF (%) | Florez et al. [35] | 0.97 |
EG-CG1: 0.75 EG-CG2: 0.75 |
EG-CG1: 9.49 EG-CG2: 10.15 |
EG-CG1: 5.0 EG-CG2: 5.2 |
EG-CG1: 5 EG-CG2: 5.2 |
|||||
| Smith et al. [38] | ≤ 0.001 | 0.55 | 0.29–0.74 | ||||||||
| Qiao et al. [40] | 0.91 | (− 7.2)–9.5 | 3.28 | ||||||||
| Graybeal J. et al. [41] | 0.221 | 0.92 | 4.3 | 2.5 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tinsley M. et al. [21] |
CG-EG1: < 0.01 CG-EG2: 0.03 |
EG1: 0.996 EG2: 0.997 |
CG-EG1: 0.90 CG-EG2: 0.89 |
EG1: 0.994–0.997 EG2: 0.996–0.998 |
CG-EG1: 4.5 CG-EG2: 4.5 |
CG-EG1: 3.4–3.5 CG-EG2: 3.4–3.5 |
|||||
| Graybeal J.; Brander F.; Tinsley M. [43] |
CG-EG1: < 0.001 CG-EG2: < 0.001 |
CG-EG1: 0.85 CG-EG2: 0.85 |
CG-EG1: 0.72 CG-EG2: 0.72 |
CG-EG1: 10.0 CG-EG2: 10.9 |
CG-EG1: 5.07 CG-EG2: 5.02 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV (%) | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Graybeal J.; Brander F.; Tinsley M. [44] |
CG-EG1: 0.799 CG-EG2: 0.799 CG-EG3: 0.544 |
EG1-EG2: 0.996 EG3: 0.990 |
CG: 27.5–30.3 EG1: 28.5–30.9 EG2: 28.6–31.0 EG3: 26.8–28.5 |
CG-EG1: 7.5 CG-EG2: 7.4 CG-EG3: 11.9 |
CG-EG1: 3.9 CG-EG2: 3.9 CG-EG3: 6.2 |
EG1-EG2: 3.1 EG3: 4.0 |
|||||
| BM (Kg) | Florez et al. [35] |
EG-CG1: 0.95 EG-CG2: 0.94 |
EG-CG1: 3.6 EG-CG2: 4.0 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Smith et al. [38] | < 0.001 | 0.68 | 0.46–0.82 | ||||||||
| FFM (kg) | Smith et al. [38] | 0.033 | 0.55 | 0.34–0.71 |
EG-CG1: 3.6 EG-CG2: 4.0 |
||||||
| Graybeal J.; Brander F.; Tinsley M. [43] | CG-EG1 0.011 |
CG-EG1: 0.96 CG-EG2: 0.96 |
CG-EG1: 0.93 CG-EG2: 0.93 |
EG1-EG2: 0.997–1.000 |
CG-EG1: 7.6 CG-EG2: 7.7 |
CG-EG1: 3.86 CG-EG2: 3.93 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV (%) | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Graybeal J.; Brander F.; Tinsley M. [44] |
CG-EG1: 0.911 CG-EG2: 0.918 CG-EG3: 0.750 |
EG1-EG2: 0.999 EG3: 0.999 |
CG: 52.8–57.0 EG1: 52.3–55.9 EG2: 52.2–55.9 EG3: 54.0–58.0 |
CG-EG1: 5.8 CG-EG2: 5.8 CG-EG3: 10.1 |
CG-EG1: 3.1 CG-EG2: 3.1 CG-EG3: 5.2 |
EG1-EG2: 1.5 EG3: 1.7 |
|||||
| WHR | Smith et al. [34] |
EG -CG1: < 0.001 EG-CG2: < 0.05—< 0.01 CG1-CG2: < 0.05—< 0.01 |
EG-CG2: 0.72–0.93 CG1-CG2: 0.78–0.95 |
EG: 2.5–6.1 CG1: 2.9–9.2 |
EG: 1–2 CG2: 0.1 |
F; 0.0169 M; 0.0363 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Choudhary et al. [37] |
F; 0.52 M; 0.61 |
F; 0.0169 M; 0.0122 |
|||||||||
| TLM (kg) | Qiao et al. [40] | 0.97 | (− 6.9)–5 | 2.54 | |||||||
| ALM (kg) | Qiao et al. [40] | 0.97 | (− 4.2)–1.8 | 1.40 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|
| McCarthy et al. [42] |
F; CG: 0.0008 EG: 0.01 M; CG: 0.002 EG: 0.02 |
F; CG: 0.74 EG: 0.79 M; CG: 0.90 EG: 0.95 |
F; CG: (− 0.40)–0.03 EG: (− 0.24)–0.16 M; CG: (− 0.23)–0.03 EG: \ (− 0.10)–0.10 |
F; CG: 1.56 EG: 1.78 M; CG: 1.53 EG: 1.50 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| FM gynoid (kg) | Qiao et al. [40] | 0.94 | (− 1.1)–1.2 | 0.51 | |||||||
| FM visceral (kg) | Qiao et al. [40] | 0.90 | (− 0.4)–0.4 | 0.37 | |||||||
| FM android(kg) | Qiao et al. [40] | 0.95 | (− 0.6)–0.8 | 0.38 | |||||||
| BSA (cm2) | Graybeal J. et al. [41] | 0.002 | 0.98 | 155–880 | 659 | ||||||
| VAT (cm3) | Graybeal J. et al. [41] | 0.91 | 304 | 151 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| BMC (cm3) | Graybeal J. et al. [41] | 0.95 | 190 | 127 | |||||||
| BH (cm) | Smith et al. [34] | < 0.05 | 0.922–0.995 | (− 1.4)-2.4 | 2 | ||||||
| Bušić et al. [36] | 0.997 | 0.995 | 0.990 |
CG: 5.09 EG: 5.27 |
|||||||
| LBM (kg) | Graybeal J. et al. [41] | = < 0.05 | 0.98 | 1.0–4.0 | 2.3 |
| Study, year | p | r | R2 | ICC | CCC | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|
| BV (cm3) | Graybeal J. et al. [41] | 0.97 | 946–7760 | 467 | ||||||
| LBC (°) | Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | 0.79 | ||||||||
| TK (°) | Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | 0.56 |
| Study, year | p | r | R2 | ICC | CCC | 95%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| TK (°) | Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | 0.56 | 0.36–0.75 | ||||||||
| NC (°) | Kandasamy; Salitkov; Schaik [39] | 0.63 | 0.44–0.80 | ||||||||
| SE (°) | Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | 0.26 | 0.07–0.52 |
| Study, year | p | r | R2 | ICC | CCC | 90%CI | LOA | RMSE | RMSE-CV | MAE | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| LPT (°) | Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | 0.09 | 0.07–0.34 | ||||||||
| FKA (°) | Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] |
L (Left); 0.40 R (Right); 0.10 |
L; (− 0.06)–0.35 R; 0.20–0.64 |
||||||||
| SP (°) | Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] |
L; 0.67 R; 0.75 |
L; 0.49–0.82 R; 0.60–0.87 |
Study risk of bias assessment
The JBI tool [45] was used to assess risk of bias. According to Xiantao Zeng et al. [46], this constitutes the most diverse set of tools, which provides a significant advantage in gaining a comprehensive overview of our three distinct studies. Observational studies were classified as descriptive or analytical (cohort [47], case–control [48], cross-sectional [49]), with cross-sectional studies [21, 33–37, 41–44] assessing exposure and outcome simultaneously, and the cohort study [40] following participants longitudinally. Experimental studies [38, 39] were appraised using the quasi-experimental checklist, reflecting their pre-post evaluation without full randomization. Three reviewers (E.V.D.H., M.S., F.V.O.) conducted independent assessments, resolving disagreements through discussion. Studies with ≥ 5 ‘yes’ responses were included; 3–4 were further assessed, ≤ 2 were excluded. Tables 9, 10 and 11 present quality assessment results by study type.
Table 9.
Risk of bias assessment: Cross-sectional
| Study | Q1 | Q2 | Q3 | Q4 | Q5 | Q6 | Q7 | Q8 | Overall appraisal |
|---|---|---|---|---|---|---|---|---|---|
| Farina et al. [33] | ✓ | ✓ | ✓ | ✓ | ✓ | ? | ✓ | ✓ | Include |
| Smith et al. [34] | ✓ | ? | ✓ | ✓ | ✓ | ? | ✓ | ✓ | Include |
| Florez et al. [35] | ✓ | ✓ | ✓ | ✓ | ✓ | ? | ✓ | ✓ | Include |
| Bušić et al. [36] | ✓ | ? | ✓ | ✓ | ✓ | ? | ✓ | ✓ | Include |
| Choudhary et al. [37] | ✓ | ✓ | ✓ | ✓ | X | ? | ✓ | ✓ | Include |
| Graybeal J. et al. [41] | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | Include |
| McCarthy et al. [42] | ✓ | ✓ | ✓ | ✓ | ✓ | ? | ✓ | ✓ | Include |
| M. Tinsley et al. [21] | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | Include |
| Graybeal J.; Brander F.; Tinsley M. [43] | ✓ | ✓ | ✓ | ✓ | ✓ | ? | ✓ | ✓ | Include |
| Graybeal J.; Brander F.; Tinsley M. [44] | ✓ | ✓ | ✓ | ✓ | ✓ | ? | ✓ | ✓ | Include |
Q1: Were the criteria for inclusion in the sample clearly defined?; Q2: Were the study subjects and the setting described in detail?; Q3: Was the exposure measured in a valid and reliable way?; Q4: Were objective, standard criteria used for measurement of the condition?; Q5: Were confounding factors identified?; Q6: Were strategies to deal with confounding factors stated?; Q7: Were the outcomes measured in a valid and reliable way?; Q8: Was appropriate statistical analysis used?
Table 10.
Risk of bias assessment: Cohort
| Study | Q1 | Q2 | Q3 | Q4 | Q5 | Q6 | Q7 | Q8 | Q9 | Q10 | Q11 | Overall appraisal |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Qiao et al. [40] | ✓ | ✓ | ✓ | ✓ | ? | ✓ | ✓ | ✓ | ? | ? | ✓ | Include |
Q1: Where the two groups similar and recruited from the same population?; Q2: Where the exposures measured similarly to assign people to both exposed and unexposed groups?; Q3: Was the exposure measured in a valis and reliable way?; Q4: Were confounding factors identified?; Q5: Were strategies to deal with confounding factors stated?; Q6: Were the groups free of the outcome at the start of the study?; Q7: Were the outcome measured in a valid and reliable way?; Q8: Was the follow up time reported and sufficient to be long enough for outcomes to occur?; Q9: Was follow up complete, and if not, were the reasons to loss to follow up described and explored?; Q10: Were strategies to address incomplete follow up utilized?; Q11: Was appropriate statistical analysis used?
Table 11.
Risk of bias assessment: quasi-experimental
| Study | Q1 | Q2 | Q3 | Q4 | Q5 | Q6 | Q7 | Q8 | Q9 | Overall appraisal |
|---|---|---|---|---|---|---|---|---|---|---|
| Smith et al. [38] | ✓ | X | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | Include |
| Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | ✓ | X | ✓ | X | ? | ✓ | ✓ | ✓ | ✓ | Include |
Q1: Is it clear in the study what is the “cause” and what is the “effect”?; Q2: Was there a control group?; Q3: Were participants included in any comparisons similar?; Q4: Were the participants included in any comparisons receiving similar treatment/care, other than the exposure or intervention of interest?; Q5: Were there multiple measurements of the outcome, both pre and post the intervention/exposure?; Q6: Were the outcomes of participants included in any comparisons measured in the same way?; Q7: Were outcomes measured in a reliable way?; Q8: Was follow-up complete and if not, were differences between groups in terms of their follow-up adequately described and analyzed?; Q9: Was appropriate statistical analysis used?
Across several studies, proportional and systematic biases were consistently observed when smartphone, tablet—or imaging-based anthropometry was compared to reference methods. Smith et al. [34] reported discrepancies between imaging approaches and traditional tape measurements, although this was not furtherss elaborated. Florez et al. [35] similarly identified proportional bias relative to both a reference model and DXA. Graybeal et al. [41] found that most outcomes demonstrated proportional bias, with the exception of BF%. In their later work, Graybeal, Brander, and Tinsley [43] observed systematic underestimation of both body fat and body mass, while their 2022 study [44] highlighted that the magnitude of proportional bias varied across sex and race. Taken together, these findings suggest that while smartphone-based 3D imaging shows promise, current applications are prone to systematic errors that limit their accuracy. Importantly, the presence of proportional bias across multiple studies indicates that measurement error is not random but influenced by participant characteristics and methodological choices. This raises concerns about generalizability and highlights the need for methodological refinements and validation across diverse populations.
Synthesis methods
Studies comparing smartphone and tablet-based body composition measurements with established reference standards were included in this systematic review. Outcome metrics included body fat percentage (BF%), fat-free mass (FFM), and fat mass (FM). When R2 or correlation coefficients (r) were missing, values were calculated to ensure consistency.
Results were presented in two main tables: an evidence table (Table 7) and a table grouped by outcome measure (Table 8). Three additional tables (Tables 9, 10 and 11) summarized the risk of bias by study type for clarity.
Due to substantial variation in outcomes and the limited number of studies per outcome, meta-analysis was not feasible. A descriptive synthesis using tables and explanatory text was therefore employed. Heterogeneity was assessed qualitatively by examining differences in study populations, smartphone models and applications, and reference standards.
Report bias assessment
Following Cochrane guidelines [50], the COSMIN checklist [51](Consensus-based Standards for the Selection of Health Measurement Instruments) was used to assess risk of bias in the thirteen included studies [21, 33–44]. This tool evaluates the methodological quality of studies on measurement properties of health-related outcomes [52] and supports assessment of reliability and validity. Specifically, Box 6 (reliability), Box 7 (measurement error), and Box 9 (hypothesis testing for construct validity) were applied. Consistent with COSMIN guidelines, the “worst score counts” principle was followed, meaning that a single ‘inadequate’ rating in a box overrides higher scores in other items.
Certainly assessment
To assess the certainty of evidence across the thirteen studies [21, 33–44], the GRADE method [53] was applied to outcomes reported in two or more studies: fat mass, body fat, body mass, fat-free mass, waist-to-hip ratio, appendicular lean mass, and height. Certainty is rated as high (★★★★), moderate (★★★), low (★★), or very low (★), reflecting confidence that the true effect aligns with the estimate.
Five factors can reduce evidence certainty: risk of bias, inconsistency, indirectness, imprecision, and publication bias (each decreasing 1–2 levels). Three factors can increase certainty: large magnitude of effect (increase 1–2 levels), all plausible confounding would reduce or strengthen the effect (increase 1 level), and dose–response gradient (decrease 1 level).
Results
Study selection
The search was performed on March 11th 2025 yielding 2172 records from five databases. A total of 332 duplicates were removed. After screening the titles and abstracts, 1798 articles were excluded. The remaining 39 articles were reviewed by the three researchers (E.V.D.H. or M.S. or F.V.O.), resulting in thirteen full-text articles [21, 33–44] being retained.
Study characteristics
The thirteen included studies comprised ten cross-sectional [21, 33–37, 41–44], one cohort [40], and two experimental studies [38, 39]. Sample sizes varied widely: cross-sectional studies included 1788 participants, the cohort study had 12,435, and the experimental studies totaled 54 participants. Both genders were represented, though proportions were not always reported, and all participants were healthy adults.
A broad range of anthropometric outcomes was assessed, with the most common being circumferences, fat mass, body fat percentage, and height. This review focuses on key outcomes including fat mass, body fat percentage, fat-free mass, and waist-to-hip ratio.
Measurement methods were generally consistent: control groups employed DXA, 4-compartment models, or SS20 scanners, while experimental groups used smartphone-based methods, including 3D scanning apps (e.g. MethreeSixty).
Risk of bias in studies
Slight variations in applying the JBI tool were observed across the three study types; details are provided in the section “Study Risk of Bias Assessment.” Each item could be rated as yes, no, unclear, or not applicable, with guidance for interpretation. Most items received a “yes” response, but question five (“Were confounding factors identified?”) showed variability, and question six (“Were strategies to address confounding factors stated?”) was often rated “unclear.” Despite these minor differences, all studies were ultimately included in the discussion.
Results of individual studies
As shown in Table 7 (“Evidence for Reliability, Concurrent Validity, and Accuracy”) and Table 8 (“Evidence for Reliability, Validity, and Accuracy”), the thirteen studies [21, 33–44] exhibit varying results. This section discusses the 23 different outcomes related to reliability, concurrent validity, and accuracy.
Fat mass (FM)
A total of five studies [33, 35, 40, 43, 44] assessed the concurrent validity and accuracy of smartphone applications across diverse populations. Sample sizes ranged from 96 to 12,435 healthy participants of different ages and genders. Control groups used DXA and 4C models, while experimental groups applied smartphone apps or portable 3D scanning. Graybeal et al. [43] employed one application designed for laboratory use and another intended for personal use, in order to demonstrate whether the latter provides reliable results in non-clinical settings.
Reliability
Overall, the ICC values in Table 8. ‘Table of evidence for reliability validity and accuracy’ suggest excellent for Farine et al. [33] (male 0.952 and female 0.974), excellent for Graybeal J.; Brander F.; Tinsley M. [43] where ICC was between EG1 and EG2 ranged from 0.997 to 1.000, good for Graybeal J.; Brander F.; Tinsley M. [44] (Amazon Halo app 0.799 and MyBVI 0.837). In Florez et al. [35] they found for the CCC a good agreement (0.86) and for Farina et al. [33] an excellent (male 0.964 and female 0.952).
Concurrent validity
All five studies [33, 35, 40, 43, 44] reported R2 values, which were generally very high (> 0.9). The only exception was Graybeal, Brander, and Tinsley [44], where R2 for the control group versus EG3 (MyBVI app) was 0.837. Across studies, the variation mainly reflected differences in reference and experimental methods: Farina et al. [33] (DXA vs. digital photography), Florez et al. [35] (DXA vs. Army Men 3D scanner), Qiao et al. [40] (Fenland study vs. 3D body-shape app), Graybeal et al. [43] (iPhone 12 Pro app vs. 4C model), and Graybeal et al. [44] (4C model vs. Amazon Halo apps). Florez et al. [35] and Graybeal et al. [43] also reported good correlations (≥ 0.85). Significant group differences were observed by Graybeal et al. [43], but not by Farina et al. [33].
RMSE values ranged from 2.5 to 5.2, indicating good predictive accuracy. Graybeal et al. [44] showed consistent performance across subsets (RMSE-CV EG1–EG2: 3.3; EG3: 4.3). Florez et al. [35] was the only study to report MAE, which was low (3.6), further supporting close agreement between predicted and actual measures. Detailed RMSE results for the four relevant studies [33, 40, 43, 44] are provided in Table 8.
Summary of evidence
Overall, anthropometric measurements in the control group versus smartphone and tablet based showed good reliability and accuracy. The validity is rather good. These findings indicate the important grow in smartphone and tablet-based measurements for fat mass.
Body fat percentage (BF%)
Seven studies [21, 35, 38, 40, 41, 43, 44] included populations of 38–184 participants with varying age and gender distributions. Experimental groups used smartphone applications [21, 38, 40, 43, 44] or a portable 3D scanner [35], while control groups employed DXA [21, 35, 38, 40] or 4C models [35, 43, 44].
Reliability
Tinsley M. et al. [21] and Graybeal J.; Brander F.; Tinsley M. [44] had an excellent ICC (30.99). Florez et al. [35] had a high agreement (CCC), contrary to Smith et al. [38] who had a moderate agreement. Tinsley M. et al. [21] scores superior with a very high agreement.
Concurrent validity
The Pearson correlation coefficient is scored excellent (30.85) in Florez et al. [35] and Graybeal J.; Brander F.; Tinsley M. [43]. The determination coefficient is in Qiao et al. [40] and Graybeal J. et al. [41] outperforms (> 0.9) Graybeal J.; Brander F.; Tinsley M. [43] and Graybeal J.; Brander F.; Tinsley M. [44] which are moderate to high (> 0.5).
Accuracy
In general, RMSE values were low, ranging from 2.5 Graybeal et al., [41] to 5.2 Florez et al., [35], indicating good predictive accuracy. Limits of agreement varied between 2.3 Qiao et al., [40] and 10.15 Florez et al., [35]. MAE values reported by Florez et al. ([35], 5.0–5.2) and Tinsley et al. [21] similarly suggest close agreement between predicted and actual measurements.
Summary of evidence
Overall, these articles have a moderate to high validity, except for Smith et al. [38]. They both found a significant difference, just as Graybeal J.; Brander F.; Tinsley M. [43] in contrary to Graybeal J. et al. [41]. The accuracy can also be assessed good.
Body mass (BM)
In total, two studies were included: one experimental [38] with 38 participants comparing DXA and Body Scan SPA, and one cross-sectional [35] with 96 participants. The latter used DXA and the 4C model as controls, while the experimental group (Army men) was assessed with 3D optical (3DO) imaging.
Reliability
Florez et al. [35] showed an excellent agreement (0.95–0.94), in contrast to Smith et al. [38], who reported moderate agreement (0.68; 95% CI 0.46–0.82). Florez et al. [35] also demonstrated high reliability and adequate validity (Table 12). The RMSE (EG–CG1: 3.6; EG–CG2: 4.0) indicated reasonable accuracy of 3DO, with slightly better alignment to the 4C model. Smith et al. [38] showed lower reliability and a statistically significant difference. For body mass, no accuracy values were given, and overall bias was rated as doubtful.Fat free mass (FFM).
Table 12.
Reporting bias the conclusion
| Study | Reliability | Validity | Accuracy | Conclusion |
|---|---|---|---|---|
| Farina et al. [33] | D | D | D | D |
| Smith et al. [34] | D | D | D | D |
| Florez et al. [35] | A | A | D | D |
| Bušić et al. [36] | VG | VG | VG | VG |
| Choudhary et al. [37] | D | D | VG | D |
| Smith et al. [38] | I | VG | D | I |
| Kandasamy G.; Salitkov J.B.; Schaik P.V. [39] | A | A | NA | A |
| Qiao et al. [40] | D | D | D | D |
| Graybeal J. et al. [41] | I | D | VG | I |
| McCarthy et al. [42] | I | A | VG | I |
| M. Tinsley et al. [21] | A | A | VG | A |
| Graybeal J.; Brander F.; Tinsley M. [43] | A | A | A | A |
| Graybeal J. et al. [44] | A | A | VG | A |
The following abbreviations were used: A: adequate; D: doubtful; I: inadequate and VG: very good
Three studies [38, 43, 44] were included: one experimental [38] and two cross-sectional [43, 44]. Participants (ages 18–75) ranged from 38 to 184. Experimental groups used smartphone-based apps: MethreeSixty (MTS) on iPhone 12 Pro (EG1) and Samsung Galaxy S21 (EG2) [43]; Amazon Halo (EG1–EG2) and MyBVI (EG3) [44]; and Body Scan SPA on iPhone 13 [38]. Control groups applied DXA [38] and the 4C model [43, 44].
Reliability
Graybeal J.; Brander F.; Tinsley M. [43] and Graybeal J.; Brander F.; Tinsley M. [44] indicates an excellent ICC (> 0.99). Contrary to Smith et al. [38] that reveals a moderate agreement (CCC = 0.55) and a broad interval for 95% CI (0.34–0.71).
Concurrent validity
Smith et al. [38] and Graybeal, Brander, and Tinsley [43] reported significant differences between control and experimental groups. R2 values were very high (> 0.9) for both Graybeal et al. [43] and Graybeal et al. [44], except for CG (4C model) versus EG3 (MyBVI app), which was high (0.75). Pearson correlations were excellent (> 0.9) for Graybeal et al. [43].
Accuracy
RMSE values for Graybeal, Brander, and Tinsley [43] and [44] ranged from 3.1 to 5.2, with the highest (5.2) for CG (4C model) versus EG3 (MyBVI app), indicating the largest deviation from actual measurements. This comparison also had the widest LOA (10.1 kg). Smith et al. [38] reported MAE of 3.6–4.0, suggesting close agreement with real measurements.
Summary of evidence
Overall, Graybeal, Brander, and Tinsley [43] and [44] demonstrated excellent reliability, very high concurrent validity, and good accuracy. Smith et al. [38] showed good accuracy but lower reliability, with overall validity considered doubtful. For fat-free mass, results differed depending on the control used, with DXA generally scoring lower than the 4C model.
Waist-to-hip ratio (WHR)
Two cross-sectional studies [34, 37] were included, with 59–550 participants. In the first [34], control groups used tape measurements and an SS20 scanner, compared to the experimental MeThreeSixty (MTS) app. In the second [37], the control group performed self-measurements, while the experimental group used MeasureNet.
Concurrent validity
The validity of both studies varies from low-moderate (> 0.5) Choudhary et al. [37] to moderate-high (> 0.7) Smith et al. [34]. Smith et al. [34] also finds a significant difference between the control groups and experimental group.
Accuracy
Smith et al. [34] showed an RMSE of 2.5–6.1 (EG) and 2.9–9.2 (GC1), indicating greater variability in CG1. The mean absolute error (MAE) indicates a lower prediction error in the female group (0.0363) compared to males (0.0169). This is also the same case for Choudhary et al. [37] (males: 0.0122 and females: 0.0169).
Summary of evidence
Both studies don’t mention anything about the reliability. Therefore, this can’t be assessed. The validity for Choudhary et al. [37] (especially the male gender) is better compared to Smith et al. [34]. Choudhary et al. [37] have a low MAE, what can be understood as a good prediction for the real measurement.
Appendicular lean mass (ALM)
Only two studies [40, 42] assessed appendicular lean mass (ALM), using DXA alongside SS20 McCarthy et al., [42] or TANITA BC-418 MA, SECA 280, and tape Qiao et al., [40]. In McCarthy et al. [42], the SS20 scanner served as a validated high-resolution 3D control to develop and test prediction models under lab conditions. The experimental group used the Me360 smartphone app to cross-validate these models, with DXA as the reference for ALM in both studies.
Concurrent validity
Qiao et al. [40] has a very strong correlation (0.97) between the control and experimental groups, just like McCarthy et al. [42] found a very strong connection (CG: 0.90 and EG: 0.95) for the male and a strong correlation for the female (CG: 0.74 and EG: 0.79).
Accuracy
The RMSE of the female experimental group in the study of McCarthy et al. [42] is the highest and indicates that there is room for improve. However, these numbers are below two and can be interpreted as a good prediction between reality and the estimation.
Summary of evidence
McCarthy et al. [42] has some intervals of the 95%CI that possess zero, therefore there is not any certainty of significant effect. The accuracy and validity are in general strong. However, this conclusion is only based on the RMSE and R2. For further research, it would be important to involve other outcome measurements.
Body height (BH)
Two studies [34, 36] addressed participants body height. In one study, the control group relied on either the SS20 scanner (EG1) or tape measurements (EG2), whereas the other applied classical anthropometry. Both studies included around 50 male and female participants and were cross-sectional in design.
Concurrent validity
The Pearson correlation coefficient finds for each study a very high correlation value (> 0.9), nonetheless only Smith et al. [34] found a significant difference between the control and experimental group.
Accuracy
Smith et al. [34] had a difference of 1 at the LOA. This can be interpretated that there is probably less variety. Bušić et al. [36] found an RMSE-CV from 5.27 for the experimental group and 5.09 for the control group. These outcomes aren’t very high; however, it is important that an exact measurement is made for the accuracy.
Summary of evidence
The two studies [36, 38] did not report reliability. Overall validity was very high, but accuracy was less clear. Smith et al. [34] showed doubtful accuracy but good LOA for body height, while Bušić et al. [36] demonstrated very good accuracy based on reporting bias. Further conclusions require additional outcomes.
Overall results
Based on the thirteen included studies [21, 33–44], smartphone-based measurements demonstrate good reliability and moderate-to-strong validity across various outcomes. Accuracy is generally high but varies by anthropometric parameter. The 4C model shows the strongest performance for fat mass and fat-free mass, indicating potential for future applications. Data on weight, height, waist-to-hip ratio, and appendicular lean mass are more limited, though validity remains high. Nevertheless, these findings should be interpreted cautiously, and definitive conclusions regarding routine use are premature.
Results of synthesis
The included studies [21, 33–44] exhibited substantial heterogeneity in populations, smartphone and tablet applications, and reference standards, leading to variable accuracy and reliability. Differences in participant characteristics (e.g., age, gender, body composition) further contributed to this variability. Multi-camera systems, such as iPhone TrueDepth and LiDAR, generally outperformed single-camera applications. TrueDepth combines infrared sensing with detailed 3D imaging, while LiDAR enhances depth capture via Time-of-Flight. Some iPhone-based systems integrate stereophotogrammetry with LiDAR or use the VCSEL of TrueDepth to generate a metric point cloud for surface reconstruction; Apple’s Object Capture API also enables 3D model creation. Comparative studies suggest accuracy differs between TrueDepth- and LiDAR-based approaches, highlighting the importance of underlying technology.
Although DXA remains the gold standard, other methods, including 4C models and traditional anthropometry, also informed validity. Bias analyses revealed systematic underestimation of body fat, and in some cases, body mass, often influenced by body composition. Thus, while smartphone-based imaging shows promise, reliability depends on algorithm calibration. Overall, 3D–4D smartphone and tablet models hold potential for body composition assessment, but further advances in algorithm refinement, machine learning, and standardized scanning protocols are needed to improve reliability and clinical applicability.
Reporting bias
The risk of bias assessment was independently conducted by two of the three reviewers (E.V.D.H., M.S., or F.V.O.). Discrepancies were resolved through discussion, with the third reviewer consulted if needed. Conclusions on the reliability, validity, and accuracy of the studies are summarized in Table 12.
Across the thirteen included studies, potential sources of bias were assessed as reported by the authors. Smith et al. [34] noted bias between imaging and tape measurements, though without further discussion. Florez et al. [35] observed proportional bias when comparing smartphone-based results with DXA. Graybeal et al. [41] found proportional bias for all outcomes except BF%, while Graybeal, Brander, and Tinsley [43] reported underestimation of body fat and mass. In a later study, the same group [44] noted proportional bias with variations by sex and race. In contrast, no significant biases were reported by Farina et al. [33], Bušić et al. [36], Choudhary et al. [37], Smith et al. [38], Kandasamy et al. [39], Qiao et al. [40], McCarthy et al. [42], and Tinsley et al. [21].
This review itself shows no risk of publication, reporting, or measurement bias. All screened articles were published, outcomes were transparently reported, and results were consistently presented in tables. Furthermore, agreement among the three reviewers minimized the risk of detection or performance bias.
Certainly of evidence
The GRADE approach [53] was applied by outcome rather than by study, as each study reported multiple outcomes (Table 13). Six outcomes were assessed: fat mass, body fat percentage, body mass, fat-free mass, waist-to-hip ratio, and appendicular lean mass. Certainty was highest for fat mass (★★★★), moderate for body mass and fat-free mass, and lowest for waist-to-hip ratio and appendicular lean mass (★).
Table 13.
Certainty of evidence: GRADE
| N° | Study design | Risk of Bias | Inconsistency | Indirectness | Imprecision | Considerations | Findings | Certainty | Importance | |
|---|---|---|---|---|---|---|---|---|---|---|
| Fat mass | ||||||||||
| 5 | Cross-sectional [33, 35, 43, 44] | Not serious | Not serious | Not serious | Not serious | None | = | ⭑⭑⭑⭑ | Critical | |
| Observational cohort [40] | Not serious | Not serious | Not serious | Not serious | None | = | ⭑⭑⭑⭑ | Critical | ||
| Body fat percentage | ||||||||||
| 7 | Cross-sectional [21, 34, 35, 41, 44] | Not serious | Not serious | Serious | Not serious | None | = | ⭑ | Critical | |
| Observational cohort [40] | Not serious | Not serious | Not serious | Not serious | None | = | ⭑⭑⭑⭑ | Critical | ||
| Body mass | ||||||||||
| 2 | Cross-sectional [35, 38] | Not serious | Not serious | Not serious | Not serious | None | = | ⭑⭑⭑ | Critical | |
| Fat free mass | ||||||||||
| 3 | Cross-sectional [38, 43, 44] | Not serious | Not serious | Not serious | Not serious | None | = | ⭑⭑⭑ | Critical | |
| Waist-hip-ratio | ||||||||||
| 2 | Cross-sectional [37, 38] | Serious | Serious | Not serious | Not serious | None | ? | ⭑ | Important | |
| Appendicular lean mass | ||||||||||
| 2 | Cross-sectional [40, 42] | Not serious | Serious | Not serious | Not serious | None | ? | ⭑⭑ | Important | |
| Body height | ||||||||||
| 2 | Cross-sectional [34, 36] | Not serious | Not serious | Serious | Not serious | None | ? | ⭑⭑ | Important | |
⭑ = very low
⭑⭑ = low
⭑⭑⭑ = moderate
⭑⭑⭑⭑ = high
All studies [21, 33–44] started with two stars as observational designs. Fat mass was upgraded to four stars due to large sample size (n = 12,955 across five studies: [33, 35, 40, 43, 44] and consistent findings, showing good reliability, accuracy, and validity.
Body fat percentage reached three stars: upgraded for large samples and consistency, but downgraded for indirectness, as some cohorts were homogeneous in age and gender (e.g., Smith et al., [34]: n = 59, mean age ~ 40; Kandasamy et al., [39]: n = 16, age 25).
Appendicular lean mass also reached three stars: upgraded for sample size but downgraded for inconsistency, as the two studies [40, 42] disagreed on gender effects.
Discussion
The aim of this review was to compare the accuracy, reliability, and validity of smartphone and tablet-based imaging with traditional anthropometric methods. Thirteen studies were included [21, 33–44], all involving healthy adult participants but with heterogeneous populations. Control groups used a variety of reference methods, most often DXA or the four-compartment (4C) model, but also field-based assessments such as the Army Body Composition Test.
Three main sources of heterogeneity were identified. First, study populations differed, with some focusing on athletes and others on general adults. Second, the imaging technology varied, as different smartphones, applications, and algorithms were used. Third, reference standards were inconsistent, ranging from DXA to 4C models and field-based tools. These differences limited comparability and prevented pooling of results. Algorithmic variability further contributed to bias, since each app uses different assumptions to process 2D or depth data. For example, some applications overestimated waist-to-hip ratio or underestimated appendicular lean mass, while one validation study [54] reported the lowest mean absolute error for smartphone-based adiposity compared to other methods.
Despite this variability, several key outcomes showed consistent promise. Correlations with gold-standard methods were high (R2 up to 0.97) [40, 41]. Fat mass, assessed in five studies [33, 35, 40, 43, 44] demonstrated the strongest agreement with DXA and 4C, supported by relatively large, combined samples and high methodological certainty. By contrast, outcomes such as waist-to-hip ratio, appendicular lean mass, and body mass (investigated in only two studies [34, 35] produced less conclusive findings.
Methodological factors contributed to these inconsistencies. While heterogeneity can increase generalizability, some studies included very homogeneous groups (e.g., Smith et al. [34]; Kandasamy et al. [39]), which limited strength. Control groups also differed, and standardizing reference methods, ideally DXA, would improve comparability. Differences in smartphone and tablet hardware played a significant role in outcome variability. Devices equipped with multi-camera systems, such as Apple’s TrueDepth and LiDAR technologies, consistently outperformed single-camera models [38, 40, 43, 44]. TrueDepth integrates infrared sensing with structured 3D imaging, enabling detailed facial and body surface capture and improving applications such as augmented reality. LiDAR, based on time-of-flight measurements, enhances depth sensing and autofocus, leading to higher image resolution and more accurate reconstruction of body planes. These technologies provide more precise surface models, thereby reducing measurement error compared to single-lens systems.
Smartphone and tablet-based anthropometric tools may play an important role at multiple levels. On the micro level, individual users could monitor fat mass or fat-free mass through standardized applications. On the meso level, athletes, coaches, and dietitians could use these tools to track body composition during training cycles, allowing for rapid adjustments in nutrition or training plans. On the macro level, potential integration into healthcare systems raises broader social, ethical, and managerial considerations. Governmental bodies and regulatory agencies will ultimately decide on approval and reimbursement, but the scientific community, clinicians and patients should guide the process to ensure safe and effective implementation. Establishing such a collaborative framework will be essential for financial support and for acceptance of smartphone and tablet-based anthropometry as a medical application in clinical practice.
This review has several limitations. No eligible studies using 4D scanning were identified, restricting conclusions to 3D methods. Many included studies had small sample sizes, reducing statistical power. Reference standards varied across studies, complicating direct comparisons. Finally, heterogeneity in populations, reference methods, and technologies, while broadening scope, limited outcome comparability and precluded meta-analysis. Larger, standardized validation studies are needed to establish the clinical utility of smartphone and tablet-based anthropometry.
Conclusion
Overall, smartphone and tablet-based body measurements show promising reliability, accuracy, and validity, particularly for fat mass. However, findings for outcomes such as waist-to-hip ratio and appendicular lean mass remain inconsistent, largely due to limited algorithm refinement, inconsistent use of validated control groups, and lack of standardized scanning protocols. Although variation exists across studies, technology is advancing rapidly. This review initially aimed to include both 3D and 4D measurements, but due to insufficient literature, the analysis was restricted to 3D methods. With continued progress in algorithm development, multi-camera systems, machine learning, and standardized protocols, smartphone and tablet-based imaging has strong potential as a practical, fast, and user-friendly tool for assessing body composition in clinical and athletic settings. Such applications may benefit individuals (micro), coaches and professionals (meso), and health care systems (macro). While current evidence is insufficient to support their standalone use, smartphone and tablet-based measurements represent a promising tool for anthropometric assessment.
Author contributions
The author contribution statement using initials for your provided author list: Sofia Scataglini → S.S. Femke Van Order → F.V.O. Emma Van Den Hoek → E.V.D.H. Mirte Stessel → M.S. Steven Truijen → S.T. Author Contributions: Conceptualization: S.S., F.V.O., E.V.D.H., M.S., S.T.; Data curation, formal analysis, investigation, methodology, and software: S.S., F.V.O., E.V.D.H., M.S.; Project administration, supervision, and resources: S.S., S.T.; Writing – original draft: S.S., F.V.O., E.V.D.H., M.S.; Writing – review and editing: S.S., S.T. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Data availability
All titles and abstracts resulting from the database search were directly imported into Rayyan for screening and management. The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. Due to licensing restrictions from the database providers, raw data (i.e., full records retrieved from proprietary databases) cannot be publicly shared.
Declarations
Ethics approval and consent to participate
Not applicable.
Consent for publication
Not applicable (as it is a systematic review).
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Ng BK, Hinton BJ, Fan B, Kanaya AM, Shepherd JA. Clinical anthropometrics and body composition from 3D whole-body surface scans. Eur J Clin Nutr. 2016;70(11):1265–70. 10.1038/ejcn.2016.109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Piqueras P, Ballester A, Durá-Gil JV, Martinez-Hervas S, Redón J, Real JT. Anthropometric indicators as a tool for diagnosis of obesity and other health risk factors: a literature review. Front Psychol. 2021. 10.3389/fpsyg.2021.631179. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Rauter S, Simenko J. Morphological asymmetries profile and the difference between low- and high-performing road cyclists using 3D scanning. Biology. 2021;10(11):1199. 10.3390/biology10111199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Metzler V, Arampatzis A, Brüggemann G. Influence of 2D and 3D body segment models on energy calculations during kinematic analysis of running. Eur J Appl Physiol. 2002;86(4):337–41. 10.1007/s00421-001-0539-3. [DOI] [PubMed] [Google Scholar]
- 5.Makowski K, Okrasa M. Application of 3D scanning and 3D printing for designing and fabricating customized half-mask facepieces: a pilot study. Work. 2019;63(1):125–35. 10.3233/wor-192913. [DOI] [PubMed] [Google Scholar]
- 6.Graybeal AJ, Brandner CF, Tinsley GM. Evaluation of automated anthropometrics produced by smartphone-based machine learning: a comparison with traditional anthropometric assessments. Br J Nutr. 2023;130(6):1077–87. 10.1017/s0007114523000090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Rumbo-Rodríguez L, Sánchez-SanSegundo M, Ferrer-Cascales R, García-D’Urso N, Hurtado-Sánchez JA, Zaragoza-Martí A. Comparison of body scanner and manual anthropometric measurements of body shape: a systematic review. Int J Environ Res Public Health. 2021;18(12):6213. 10.3390/ijerph18126213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Van Gils RHJ, Wauben LSGL, Helder OK. Body size measuring techniques enabling stress-free growth monitoring of extreme preterm infants inside incubators: a systematic review. PLoS ONE. 2022;17(4):e0267285. 10.1371/journal.pone.0267285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jamali AA, Deuel C, Perreira A, Salgado CJ, Hunter JC, Strong EB. Linear and angular measurements of computer-generated models: are they accurate, valid, and reliable? Comput Aided Surg. 2007;12(5):278–85. 10.3109/10929080701680265. [DOI] [PubMed] [Google Scholar]
- 10.Robinette KM, Daanen HA. Precision of the CAESAR scan-extracted measurements. Appl Ergon. 2005;37(3):259–65. 10.1016/j.apergo.2005.07.009. [DOI] [PubMed] [Google Scholar]
- 11.DeFroda SF, Thigpen CA, Kriz PK. Two-dimensional video analysis of youth and adolescent pitching biomechanics: a tool for the common athlete. Curr Sports Med Rep. 2016;15(5):350–8. 10.1249/jsr.0000000000000295. [DOI] [PubMed] [Google Scholar]
- 12.Munro A, Herrington L, Carolan M. Reliability of 2-dimensional video assessment of frontal-plane dynamic knee valgus during common athletic screening tasks. J Sport Rehabil. 2012;21(1):7–11. 10.1123/jsr.21.1.7. [DOI] [PubMed] [Google Scholar]
- 13.Bartol K, Bojanić D, Petković T, Pribanić T. A review of body measurement using 3D scanning. IEEE Access. 2021;9:67281–301. [Google Scholar]
- 14.Tian IY, Ng BK, Wong MC, Kennedy S, Hwaung P, Kelly N, et al. Predicting 3D body shape and body composition from conventional 2D photography. Med Phys. 2020;47(12):6232–45. 10.1002/mp.14492. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Tinsley GM, Rodriguez C, Siedler MR, Tinoco E, White SJ, LaValle C, et al. Mobile phone applications for 3-dimensional scanning and digital anthropometry: a precision comparison with traditional scanners. Eur J Clin Nutr. 2024;78(6):509–14. 10.1038/s41430-024-01424-w. [DOI] [PubMed] [Google Scholar]
- 16.Alemany S, Uriel J, Ballester A, Parrilla E, Instituto de Biomecánica de Valencia. Universitat Politècnica de València. Three-dimensional body shape modeling and posturography. In: Instituto de Biomecánica de Valencia. Universitat Politècnica de València. Amsterdam: Elsevier Inc; 2019. [Google Scholar]
- 17.Löffler-Wirth H, Willscher E, Ahnert P, Wirkner K, Engel C, Loeffler M, et al. Novel anthropometry based on 3D-bodyscans applied to a large population based cohort. PLoS ONE. 2016;11(7):e0159887. 10.1371/journal.pone.0159887. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Papadopoulou AK, Di Santo F, Antonarakis GS, Ghislanzoni LH. The validity of a smartphone-based method for acquiring 3D images of the face. J Clin Med. 2024;13(21):6362. 10.3390/jcm13216362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Quispe-Enriquez OC, Valero-Lanzuela JJ, Lerma JL. Smartphone photogrammetric assessment for head measurements. Sensors. 2023;23(21):9008. 10.3390/s23219008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Trujillo-Jiménez MA, Navarro P, Pazos B, Morales L, Ramallo V, Paschetta C, et al. Body2vec: 3D point cloud reconstruction for precise anthropometry with handheld devices. J Imaging. 2020;6(9):94. 10.3390/jimaging6090094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Tinsley GM, Rodriguez C, Florez CM, Siedler MR, Tinoco E, McCarthy C, et al. Smartphone three-dimensional imaging for body composition assessment using non-rigid avatar reconstruction. Front Med. 2024. 10.3389/fmed.2024.1485450. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Tinsley GM, Harty PS, Siedler MR, Stratton MT, Rodriguez C. Improved precision of 3-dimensional optical imaging for anthropometric measurement using non-rigid avatar reconstruction and parameterized body model fitting. Clin Nutr Open Sci. 2023;50:40–5. 10.1016/j.nutos.2023.07.002. [Google Scholar]
- 23.Lamb K. Test-retest reliability in quantitative physical education research: a commentary. Eur Phys Educ Rev. 1998;4(2):145–52. 10.1177/1356336x9800400205. [Google Scholar]
- 24.Meeus M, Gebruers N. Health literacy from reference to review; ACCO: Seattle, WA, USA. 2016.
- 25.Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. Updating guidance for reporting systematic reviews: development of the PRISMA 2020 statement. J Clin Epidemiol. 2021;134:103–12. 10.1016/j.jclinepi.2021.02.003. [DOI] [PubMed] [Google Scholar]
- 26.EndNote. EndNote—the best citation & reference Management tool. 2025. https://endnote.com/.
- 27.Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. Rayyan—a web and mobile app for systematic reviews. Syst Rev. 2016. 10.1186/s13643-016-0384. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.https://pubmed.ncbi.nlm.nih.gov
- 29.https://www.webofknowledge.com/
- 30.https://www.nlm.nih.gov/medline/medline_overview.html
- 31.https://training-cochraneorg.translate.goog/handbook/current?_x_tr_sl=en&_x_tr_tl=nl&_x_tr_hl=nl&_x_tr_pto=sc
- 32.https://www.sciencedirect.com
- 33.Farina G, Spataro F, De Lorenzo A, Lukaski H. A smartphone application for personal assessments of body composition and phenotyping. Sensors. 2016;16(12):2163. 10.3390/s16122163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Smith B, McCarthy C, Dechenaud ME, Wong MC, Shepherd J, Heymsfield SB. Anthropometric evaluation of a 3D scanning mobile application. Obesity. 2022;30(6):1181–8. 10.1002/oby.23434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Florez CM, Rodriguez C, Siedler MR, Tinoco E, Tinsley GM. Body composition estimation from mobile phone three-dimensional imaging: evaluation of the USA army one-site method. Br J Nutr. 2024. 10.1017/s0007114524002216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Bušić, A., Bušić, J., Coleman, J., & Šimenko, J. (2020). Comparison of Manual Anthropometry and a Mobile Digital Anthropometric System. 8th international conference on sport sciences research and technology support icSports 2020 Authors: Anita Busic BodyRecog Metrics Josip Bušić John Coleman Jožef Šimenko University Of Ljubljana. 10.5220/0010178201090115
- 37.Choudhary S, Iyer G, Smith BM, Li J, Sippel M, Criminisi A, et al. Development and validation of an accurate smartphone application for measuring waist-to-hip circumference ratio. NPJ Digit Med. 2023. 10.1038/s41746-023-00909-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Smith MK, Staynor JMD, El-Sallam A, Ebert JR, Ackland TR. Longitudinal concordance of body composition and anthropometric assessment by a novel smartphone application across a 12-week self-managed weight loss intervention. Br J Nutr. 2023;130(7):1260–6. 10.1017/s0007114523000259. [DOI] [PubMed] [Google Scholar]
- 39.Kandasamy G, Bettany-Saltikov J, Van Schaik P. Measurement of three-dimensional back shape of normal adults using a novel three-dimensional imaging mobile surface topography system (MSTS): an intra- and inter-rater reliability study. Healthcare. 2023;11(23):3099. 10.3390/healthcare11233099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Qiao C, De Lucia Rolfe E, Mak E, Sengupta A, Powell R, Watson LPE, et al. Prediction of total and regional body composition from 3D body shape. NPJ Digit Med. 2024. 10.1038/s41746-024-01289-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Graybeal AJ, Brandner CF, Compton AT, Swafford SH, Henderson A, Aultman R, et al. Smartphone derived anthropometrics: agreement between a commercially available smartphone application and its parent application intended for use at point-of-care. Clin Nutr ESPEN. 2023;59:107–12. 10.1016/j.clnesp.2023.11.021. [DOI] [PubMed] [Google Scholar]
- 42.McCarthy C, Tinsley GM, Yang S, Irving BA, Wong MC, Bennett JP, et al. Smartphone prediction of skeletal muscle mass: model development and validation in adults. Am J Clin Nutr. 2023;117(4):794–801. 10.1016/j.ajcnut.2023.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Graybeal AJ, Brandner CF, Tinsley GM. Validity and reliability of a mobile digital imaging analysis trained by a four-compartment model. J Hum Nutr Diet. 2023;36(3):905–11. 10.1111/jhn.13113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Graybeal AJ, Brandner CF, Tinsley GM. Visual body composition assessment methods: A 4-compartment model comparison of smartphone-based artificial intelligence for body composition estimation in healthy adults. Clin Nutr. 2022;41(11):2464–72. 10.1016/j.clnu.2022.09.014. [DOI] [PubMed] [Google Scholar]
- 45.Home page . JBI. (n.d.). https://jbi.global/.
- 46.Zeng X, Zhang Y, Kwong JS, Zhang C, Li S, Sun F, et al. The methodological quality assessment tools for preclinical and clinical studies, systematic review and meta-analysis, and clinical practice guideline: a systematic review. J Evid Based Med. 2015;8(1):2–10. 10.1111/jebm.12141. [DOI] [PubMed] [Google Scholar]
- 47.Checklist for cohort studies. In JBI systematic reviews. JBI. 2020. https://synthesismanual.jbi.global.
- 48.Checklist for case control studies. In JBI systematic reviews. JBI. 2020. https://synthesismanual.jbi.global.
- 49.Checklist for analytical cross sectional studies. In JBI Systematic Reviews. JBI. 2020. https://synthesismanual.jbi.global.
- 50.Cochrane Handbook for Systematic Reviews of Interventions. (n.d.). Cochrane Training. https://training.cochrane.org/handbook.
- 51.COSMIN. COSMIN-Improving the selection of outcome measurement instruments. 2025. https://www.cosmin.nl/.
- 52.Mokkink LB, Terwee CB, Knol DL, Stratford PW, Alonso J, Patrick DL, et al. The COSMIN checklist for evaluating the methodological quality of studies on measurement properties: a clarification of its content. BMC Med Res Methodol. 2010. 10.1186/1471-2288-10-22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.GRADE handbook. (n.d.). https://gdt.gradepro.org/app/handbook/handbook.html.
- 54.Majmudar MD, Chandra S, Yakkala K, Kennedy S, Agrawal A, Sippel M, et al. Smartphone camera based assessment of adiposity: a validation study. NPJ Digit Med. 2022. 10.1038/s41746-022-00628-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Five databases were searched: PubMed [28] (Table 2), Web of Science [29] (Table 3), Medline [30] (Table 4), Cochrane [31] (Table 5), and ScienceDirect [32] (Table 6). In PubMed, MeSH terms were combined with free keywords, using truncation to broaden the search. The strategy was first developed for PubMed (Table 2) and adapted for other databases. The final search was conducted on March11, 2025.
Table 2.
Database Search Strategy in PubMed
| Database: PubMed | Search strategy |
|---|---|
| P | / |
| I | ((smartphone app) OR (iPhone app) OR mobile OR (smartphone application*) OR (iPhone application*) OR (smartphone-based imaging) OR ("Mobile Applications"[Mesh]) OR (mobile applications)) |
| C | (scanning OR imaging OR photogrammetry OR infrared OR (Depth camera)) |
| O | ((anthro*) OR (metric*) OR (meas*)) AND ((Concurrent Validity) OR Validity OR reliability OR accuracy OR (landmark position error) OR (Reproducibility of results) OR comparison OR (clustering high-dimensional data) OR ("Reproducibility of Results"[Mesh]) OR (reproducibility of results)) AND (3D OR ("Imaging, Three-Dimensional"[Mesh]) OR (imaging, three-dimensional) OR 4D OR (digital human modeling) OR dynamic OR DHM OR (body shape) OR (geometric*)) |
P: population; I: intervention; C: comparison; O: outcome
Table 3.
Database Search Strategy in Web of Science
| Database: web of science | Search strategy |
|---|---|
| P | / |
| I | ALL = ((smartphone app) OR (iPhone app) OR mobile OR (smartphone application*) OR (iPhone application*) OR (smartphone-based imaging)) |
| C | ALL = (scanning OR imaging OR photogrammetry OR infrared OR (Depth camera)) |
| O | ALL = ((anthro*) OR (metric*) OR (meas*)) AND ALL = ((Concurrent Validity) OR Validity OR reliability OR accuracy OR (landmark position error) OR correlation OR (Reproducibility of results) OR comparison OR (clustering high-dimensional data)) AND ALL = (3D OR 4D OR (digital human modeling) OR dynamic OR DHM OR (body shape) OR (geometric*)) |
P: population; I: intervention; C: comparison; O: outcome
Table 4.
Database Search Strategy in Medline
| Database: medline | Search strategy |
|---|---|
| P | / |
| I | ((smartphone app) OR (iPhone app) OR mobile OR (smartphone application*) OR (iPhone application*) OR (smartphone-based imaging) OR (mobile applications)) |
| C | (scanning OR imaging OR photogrammetry OR infrared OR (Depth camera)) |
| O | ((anthro*) OR (metric*) OR (meas*)) AND ((Concurrent Validity) OR Validity OR reliability OR accuracy OR (landmark position error) OR (Reproducibility of results) OR comparison OR (clustering high-dimensional data) OR (reproducibility of results)) AND (3D OR OR (imaging, three-dimensional) OR 4D OR (digital human modeling) OR dynamic OR DHM OR (body shape) OR (geometric*)) |
P: population; I: intervention; C: comparison; O: outcome
Table 5.
Database Search Strategy in Cochrane
| Database: cochrane | Search strategy |
|---|---|
| P | / |
| I | mobile |
| C | imaging |
| O | anthro* AND meas* |
P: population; I: intervention; C: comparison; O: outcome
Table 6.
Database Search Strategy in ScienceDirect
| Database: ScienceDirect | Search strategy |
|---|---|
| P | / |
| I | mobile application |
| C | imaging |
| O | anthropometric AND measurement AND results AND 3D |
P: population; I: intervention; C: comparison; O: outcome
All titles and abstracts resulting from the database search were directly imported into Rayyan for screening and management. The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. Due to licensing restrictions from the database providers, raw data (i.e., full records retrieved from proprietary databases) cannot be publicly shared.