Computed Tomography-based Craniometric Sex Determination in Forensic Anthropology: A Systematic Review

May combine two claims.

Cranial base length was the most frequently reported and most consistent sexually dimorphic cranial parameter, appearing in approximately 72% of studies.

“Across the reviewed studies, cranial base length emerged as the most frequently reported and most consistent sexually dimorphic cranial parameter. This variable appeared in approximately 72% of studies and contributed substantially to improved classification accuracy.”

May combine two claims.

Clinical CT was the most commonly used modality, with reported classification accuracies 82.1%-93.6%.

“Clinical CT represented the most commonly used modality and accounted for the majority of studies. Reported classification accuracies for CT-based investigations ranged from 82.1% to 93.6%, depending on the cranial variables and statistical models used.”

May combine two claims.

Most included studies reported balanced or near-balanced sex distributions, but Adegbite et al. (2025) included male-only samples.

“Most investigations reported balanced or near-balanced sex distributions; however, a few studies, such as Adegbite et al. (2025),[] included male-only samples, which should be interpreted cautiously in the context of sex determination.”

• cites Adegbite et al. (2025) sourceNot checked yet

  Checked before — its earlier verdict is held back until a new check runs.

May combine two claims.

The Northwest Indian CT-based study reported the highest classification accuracy (93.6%) among Indian datasets and identified bi-infraorbital foramen breadth combined with cranial base length as highly discriminatory.

“The Northwest Indian CT-based study reported the highest classification accuracy (93.6%) among Indian datasets and ranked within the upper quartile of global CT-based studies. This investigation identified bi-infraorbital foramen breadth in combination with cranial base length as a highly discriminatory parameter.”

May combine two claims.

Advanced phase (2015-2020) had 17 studies, with classification accuracies 83.1%-94.7%, increasingly using stepwise discriminant models.

“Seventeen studies were published between 2015 and 2020. These investigations were characterized by larger sample sizes, improved landmark standardization, and population-specific statistical models. Reported classification accuracies ranged from 83.1% to 94.7%, with stepwise discriminant models increasingly used to optimize variable selection.”

May combine two claims.

CBCT-based studies reported classification accuracies 87.9%-91.2%, suggesting sufficient spatial resolution for craniofacial sex determination when appropriate variables are selected.

“Despite a smaller field of view compared with clinical CT, CBCT studies reported classification accuracies ranging from 87.9% to 91.2%, suggesting that CBCT provides sufficient spatial resolution for craniofacial sex determination when appropriate anatomical variables are selected.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

After duplicate removal, 312 records were screened, and 122 articles were retained for full-text assessment.

“After removal of duplicate records (n = 54), 312 records were screened based on titles and abstracts, of which 122 articles were retained for full-text assessment.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Sample sizes across included studies ranged from 78 to 300 individuals.

“Sample sizes varied considerably across studies, ranging from 78 to 300 individuals.”

May combine two claims.

Multivariate DFA significantly improved predictive performance, with most studies reporting classification accuracies between 88% and 92%.

“Multivariate DFA significantly improved predictive performance, with most studies reporting classification accuracies between 88% and 92%.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A total of 366 records were initially identified across all sources.

“A total of 366 records were initially identified across all sources, including electronic databases and additional reference list screening.”

May combine two claims.

Mandibular measurements were examined in a smaller number of studies and generally served as supplementary predictors rather than primary discriminating variables.

“Mandibular measurements were examined in a smaller number of studies and generally served as supplementary predictors rather than primary discriminating variables.”

May combine two claims.

Intermediate phase (2010-2015) had 22 studies, with classification accuracies 85.0%-93.4%, particularly in East Asian populations.

“Between 2010 and 2015, 22 studies were published, marking a period of methodological expansion. During this phase, the adoption of MSCT increased, and statistical modeling became more sophisticated. Classification accuracies improved to 85.0%–93.4%, particularly in East Asian populations where large clinical CT datasets were available.”

May combine two claims.

Stepwise DFA consistently demonstrated the highest performance across populations, frequently exceeding 90% and approaching 95%-96% in recent studies.

“Stepwise DFA consistently demonstrated the highest performance across populations. In recent studies, stepwise models frequently achieved classification accuracies exceeding 90%, with several investigations reporting accuracies approaching 95%–96%.”

May combine two claims.

Research concentration is in East Asia and Europe, with limited representation in Africa and parts of South America.

“Figure 2 illustrates the global distribution of CT-based craniofacial sex determination studies, showing a concentration of research in East Asia and Europe, with limited representation in Africa and parts of South America, highlighting the need for increased studies in underrepresented populations to improve global forensic applicability.”

• AssumptionRests on an assumption: The 65 included studies are representative of the global literature on CT-based craniometric sex determination.

May combine two claims.

Early phase (2005-2009) had 17 studies, with reported classification accuracies 85.1%-89.2%, mostly using multivariate DFA.

“Seventeen studies were published during the early phase of CT-based research. These investigations primarily focused on validating CT-derived cranial measurements against traditional osteometric methods and assessing measurement reproducibility. Reported classification accuracies ranged from 85.1% to 89.2%, with most studies employing multivariate DFA.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

57 studies were excluded at full-text stage: lack of quantitative data (n=18), non-CT imaging (n=14), absence of sex classification analysis (n=11), and methodological limitations (n=14).

“A total of 57 studies were excluded at the full-text stage for the following reasons: lack of quantitative data (n = 18), use of non-CT imaging modalities (n = 14), absence of sex classification analysis (n = 11), and methodological limitations (n = 14).”

May combine two claims.

Other frequently used variables included bizygomatic breadth, foramen magnum dimensions, facial height, orbital dimensions, and nasal measurements, with moderate-to-strong discriminatory power that varied among populations.

“Other frequently used variables included bizygomatic breadth, foramen magnum dimensions, facial height, orbital dimensions, and nasal measurements. While these variables demonstrated moderate-to-strong discriminatory power, their predictive value often varied among populations.”

May combine two claims.

MSCT-based studies were fewer but consistently demonstrated high classification performance, with Japanese investigations reporting accuracies 93.4%-94.7%.

“MSCT-based studies were fewer but consistently demonstrated high classification performance. Several Japanese investigations using MSCT reported some of the highest accuracies in the literature, ranging from 93.4% to 94.7%.”

May combine two claims.

The included studies focused exclusively on adult populations (≥18 years) and reported quantitative craniofacial measurements used for sex determination.

“The investigations focused exclusively on adult populations (≥18 years) and reported quantitative craniofacial measurements used for sex determination.”

May combine two claims.

Limitations across included studies included lack of external validation, limited reporting of intra- and inter-observer reliability, and potential sample bias in region-specific datasets.

“However, certain limitations were identified, including a lack of external validation, limited reporting of intra- and inter-observer reliability, and potential sample bias in region-specific datasets.”

May combine two claims.

European studies frequently reported strong discriminatory power for bizygomatic breadth and facial height, whereas African and Middle Eastern studies emphasized nasal dimensions and cranial vault indices as important predictors of sex.

“European studies frequently reported strong discriminatory power for bizygomatic breadth and facial height, whereas studies conducted in African and Middle Eastern populations emphasized nasal dimensions and cranial vault indices as important predictors of sex.[,,,,]”

May combine two claims.

Clinical CT was the most widely used imaging modality in the reviewed studies due to its availability in hospital settings and its ability to generate high-resolution three-dimensional images of cranial structures.

“Clinical CT was the most widely used imaging modality due to its availability in hospital settings and its ability to generate high-resolution three-dimensional images of cranial structures.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The PRISMA 2020 flow diagram was adapted from Page et al. (2021) and reproduced under CC BY 4.0.

“The PRISMA 2020 flow diagram was adapted from the PRISMA statement published by Page et al. (2021)[] and reproduced under the Creative Commons Attribution License (CC BY 4.0).”

• cites Page et al. (2021) sourceNot checked yet

  Queued — the background check reaches it, usually within a day.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Only a limited number of machine learning investigations met the inclusion criteria and are not extensively represented in Table 1.

“Although machine learning methods are mentioned in several investigations, only a limited number met the inclusion criteria and are therefore not extensively represented in Table 1, which primarily summarizes studies employing classical statistical approaches.”

Probably not a substantial claim: may not pass S4 (based).

CT imaging allows visualization of internal cranial structures that may remain intact even when external morphology is damaged.

“CT imaging allows visualization of internal cranial structures that may remain intact even when external morphology is damaged.[,,,]”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Studies utilizing MSCT frequently reported some of the highest classification accuracies in the literature.

“Studies utilizing MSCT frequently reported some of the highest classification accuracies in the literature.[,,,,]”

May combine two claims.

Indian population studies exhibit greater variability in classification accuracy than other populations, reflecting the high genetic and anthropological diversity of the subcontinent.

“In contrast, Indian population studies exhibit greater variability, reflecting the high genetic and anthropological diversity of the subcontinent.[,,,]”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Recent Indian investigations have identified region-specific craniofacial parameters, such as bi-infraorbital foramen breadth, that significantly enhance classification accuracy.

“Notably, recent Indian investigations have identified region-specific craniofacial parameters, such as bi-infraorbital foramen breadth, which significantly enhance classification accuracy.[]”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The global characteristics of the 65 studies are summarized in Table 1, and remaining studies grouped regionally in Table 2.

“The global characteristics of the 65 CT-based craniofacial sex determination studies included in this systematic review, including population origin, imaging modality, sample size, statistical approach, and reported classification accuracy, are summarized in Table 1. The remaining studies were grouped regionally due to methodological similarity and are summarized in Table 2.”

May combine two claims.

East Asian and European populations frequently reported higher classification accuracy (approximately 90%-95%), whereas South Asian, Middle Eastern, and African studies showed wider variability (approximately 83%-93%).

“Regionwise analysis indicated that East Asian and European populations frequently reported higher classification accuracy (approximately 90%–95%), whereas studies from South Asia, the Middle East, and Africa demonstrated slightly wider variability (approximately 83%–93%), reflecting population-specific differences and sample heterogeneity.”

• AssumptionRests on an assumption: The included studies' reported classification accuracies are sufficiently comparable to permit a structured comparative synthesis despite heterogeneity.

May combine two claims.

Population-specific reference standards are important for craniofacial sex determination, and applying discriminant functions developed for one population to individuals from another population reduces classification accuracy.

“One of the most consistent findings across the reviewed literature is the importance of population-specific reference standards for craniofacial sex determination. Numerous studies reported reduced classification accuracy when discriminant functions developed for one population were applied to individuals from another population.[,,,,]”

May combine two claims.

MSCT-based studies generally reported the highest accuracy (approximately 93%-95%), followed by clinical CT (82%-93%) and CBCT (approximately 87%-91%).

“With respect to imaging modality, MSCT-based studies generally reported the highest accuracy (approximately 93%–95%), followed by clinical CT (82%–93%) and CBCT (approximately 87%–91%).”

• AssumptionRests on an assumption: The included studies' reported classification accuracies are sufficiently comparable to permit a structured comparative synthesis despite heterogeneity.

May combine two claims.

Stepwise discriminant function analysis enhances predictive performance by identifying the most informative subset of cranial variables while eliminating redundant predictors, and many of the highest reported classification accuracies were achieved using stepwise models.

“Stepwise DFA further enhances predictive performance by identifying the most informative subset of cranial variables while eliminating redundant predictors. Many of the highest reported classification accuracies in the reviewed literature were achieved using stepwise models.[,,,,]”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Studies incorporating cross-validation or independent testing samples generally demonstrate greater reliability and reduce the risk of overfitting.

“Studies incorporating cross-validation or independent testing samples generally demonstrate greater reliability and reduce the risk of overfitting, which is essential for ensuring forensic applicability in medicolegal contexts.[,,,,]”

May combine two claims.

More recent CT studies frequently achieve classification accuracy exceeding 90%, with several investigations reporting values approaching 95%.

“whereas more recent studies frequently achieve accuracies exceeding 90%, with several investigations reporting values approaching 95%.”

• Also stated in: § Conclusion · boosted

  “Advances in imaging technology and statistical modeling have significantly improved classification accuracy, with recent studies consistently exceeding 90%.”

May combine two claims.

Univariate statistical approaches were useful for identifying sexually dimorphic cranial traits but generally produced lower classification accuracy because they evaluate each variable independently.

“Univariate statistical approaches were useful for identifying sexually dimorphic cranial traits but generally produced lower classification accuracy because they evaluate each variable independently.”

May combine two claims.

CT-derived craniofacial measurements provide a reliable and reproducible method for sex estimation, particularly in forensic contexts where pelvic remains are absent, damaged, or unavailable.

“The analysis of 65 eligible studies demonstrates that CT-derived craniofacial measurements provide a reliable and reproducible method for sex estimation, particularly in forensic contexts where pelvic remains are absent, damaged, or unavailable.”

• AssumptionRests on an assumption: In forensic contexts, pelvic remains are sometimes absent, damaged, or unavailable, creating a need for alternative sex estimation methods.

Classified as probably not a substantial claim.

Geographic representation in CT-based craniofacial research remains uneven, with many global populations still underrepresented.

“Geographic representation in CT-based craniofacial research remains uneven, with many global populations still underrepresented.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Angular measurements were less frequently used but provided complementary information regarding cranial shape and orientation.

“Angular measurements were less frequently used but provided complementary information regarding cranial shape and orientation.”

May combine two claims.

Several studies reported CBCT classification accuracies between 87% and 91%, indicating that CBCT can be effectively used for craniofacial sex determination when appropriate variables are selected.

“Although CBCT typically provides a smaller field of view than clinical CT, several studies reported classification accuracies between 87% and 91%, indicating that CBCT can be effectively used for craniofacial sex determination when appropriate variables are selected.[,,,]”

May combine two claims.

In Indian cohorts, earlier investigations using conventional multivariate models reported moderate accuracy, whereas more recent studies using stepwise discriminant analysis and larger CT datasets achieved substantially improved performance.

“Earlier investigations in Indian cohorts reported moderate accuracy using conventional multivariate models, whereas more recent studies incorporating stepwise discriminant analysis and larger CT datasets have achieved substantially improved performance.[,,]”

Probably not a substantial claim: may not pass S3 (load-bearing).

Machine learning methods remain less frequently used than classical discriminant analyses in forensic morphometrics.

“Although machine learning methods remain less frequently used than classical discriminant analyses, their application in forensic morphometrics is increasing and represents a promising direction for future research.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Several post-2020 studies have reported high classification accuracy using CT-derived datasets, often exceeding 90%, by leveraging automated feature extraction and nonlinear modeling of craniofacial morphology.

“Several post-2020 studies have reported high classification accuracy using CT-derived datasets, often exceeding 90%, by leveraging automated feature extraction and nonlinear modeling of craniofacial morphology.”

Probably not a substantial claim: may not pass S4 (based).

CT-based measurements can be obtained even in cases involving fragmented or burned skeletal remains, which are common in mass disasters and high-impact forensic scenarios.

“Second, CT-based measurements can be obtained even in cases involving fragmented or burned skeletal remains, which are common in mass disasters and high-impact forensic scenarios.”

May combine two claims.

Indian population-specific CT studies included in the review demonstrate classification accuracies ranging from approximately 83.1% to 93.6%, comparable to those reported in European and East Asian populations.

“Indian population-specific studies included in this review demonstrate classification accuracies ranging from approximately 83.1% to 93.6%, which are comparable to those reported in European and East Asian populations.[,,,,,]”

May combine two claims.

Cranial base length was included in a majority of high-performing classification models and demonstrated strong discriminatory power across diverse populations.

“This parameter was included in a majority of high-performing classification models and demonstrated strong discriminatory power across diverse populations.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

31 investigations were individually summarized, while the remaining studies were grouped regionally.

“Among the included studies, 31 investigations were individually summarized, while the remaining studies were grouped regionally because of methodological similarities and comparable analytical approaches.”

Classified as probably not a substantial claim. May combine two claims.

AI-based models remain limited in forensic applicability due to lack of external validation, dependence on large annotated datasets, limited interpretability of model outputs, and absence of standardized protocols across populations.

“However, despite their promising performance, AI-based models remain limited in forensic applicability due to several challenges, including a lack of external validation, dependence on large annotated datasets, limited interpretability of model outputs, and the absence of standardized protocols across populations.”

May combine two claims.

Bizygomatic breadth was one of the most frequently used and most reliable facial parameters in sex classification models.

“Among these, bizygomatic breadth was one of the most frequently used and most reliable parameters.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

CT imaging provides high-resolution, three-dimensional visualization of cranial structures, allowing precise landmark identification and measurement of external and internal cranial dimensions.

“CT imaging provides high-resolution, three-dimensional visualization of cranial structures, allowing precise identification of anatomical landmarks and accurate measurement of both external and internal cranial dimensions.”

Classified as probably not a substantial claim.

CT-based measurements can be obtained without physical manipulation of skeletal remains, preserving evidentiary integrity.

“Unlike traditional osteometry, CT-based measurements can be obtained without physical manipulation of skeletal remains, thereby preserving evidentiary integrity while enabling detailed morphological analysis.[]”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The review protocol was not prospectively registered in PROSPERO, but a predefined structured protocol was followed to minimize selection bias and enhance methodological transparency.

“Although the review protocol was not prospectively registered in PROSPERO, a predefined structured protocol was followed to minimize selection bias and enhance methodological transparency.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A structured protocol outlining objectives, eligibility criteria, search strategy, and data extraction procedures was developed before the review process.

“A structured protocol outlining the research objectives, eligibility criteria, search strategy, and data extraction procedures was developed before initiating the review process.”

Classified as probably not a substantial claim.

The methodological quality of included studies was assessed using a modified version of the Joanna Briggs Institute Critical Appraisal Checklist for analytical cross-sectional studies.

“The methodological quality of included studies was assessed using a modified version of the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for analytical cross-sectional studies.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The review followed PRISMA 2020 guidelines.

“The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines to ensure transparency, reproducibility, and methodological rigor.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

CT datasets can be retrospectively analyzed from clinical archives, allowing large and demographically diverse populations to be examined without additional radiation.

“Furthermore, CT datasets can be retrospectively analyzed from clinical imaging archives, allowing researchers to examine large and demographically diverse populations without exposing individuals to additional radiation.[]”

Classified as probably not a substantial claim.

Study selection and data extraction were independently performed by two reviewers.

“Study selection and data extraction were independently performed by two reviewers to minimize selection bias and ensure methodological rigor.”

May combine two claims.

The pelvis is the most reliable anatomical indicator of sex, with classification accuracies exceeding 90%-95% when well preserved.

“Among skeletal structures, the pelvis is widely recognized as the most reliable anatomical indicator of sex because of its pronounced sexual dimorphism associated with obstetric adaptations. Pelvic morphology often enables classification accuracies exceeding 90%–95% when well preserved.[]”

May combine two claims.

CT-based craniofacial morphometry has evolved into a robust but context-dependent method that requires population-specific validation and methodological standardization.

“This systematic review demonstrates that CT-based craniofacial morphometry has evolved into a robust but context-dependent method that requires population-specific validation and methodological standardization.”

Classified as probably not a substantial claim. May combine two claims.

The strong sexual dimorphism observed in cranial base length is related to its developmental stability: the cranial base undergoes early growth, is largely completed during adolescence, and its morphology is primarily determined by endocranial growth and hormonal influences during puberty.

“The biological explanation for the strong sexual dimorphism observed in cranial base length is related to its developmental stability. The cranial base undergoes early growth and is largely completed during adolescence. Unlike facial dimensions, which may be influenced by functional factors such as mastication or environmental adaptation, cranial base morphology is primarily determined by endocran”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

MSCT provides thinner slice acquisition and improved spatial resolution, enhancing landmark identification and measurement precision.

“MSCT technology provides thinner slice acquisition and improved spatial resolution, which enhances landmark identification and measurement precision.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Variability in measurement protocols introduces potential observer-related bias.

“In addition, variability in measurement protocols introduces potential observer-related bias.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Machine learning algorithms allow automated feature selection and nonlinear modeling of complex morphometric relationships, potentially improving predictive performance compared with traditional statistical techniques.

“These computational approaches allow automated feature selection and nonlinear modeling of complex morphometric relationships, potentially improving predictive performance compared with traditional statistical techniques.”

May combine two claims.

Many studies reported high classification accuracy without external validation, raising concerns regarding overfitting and limited generalizability in forensic casework.

“Many studies reported high classification accuracy without external validation, raising concerns regarding overfitting and limited generalizability in forensic casework.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

The introduction of MSCT allowed the acquisition of thin-slice images with improved spatial resolution and reduced measurement error.

“The introduction of MSCT allowed the acquisition of thin-slice images with improved spatial resolution and reduced measurement error.”

May combine two claims.

Many included studies lacked external validation and relied on homogeneous population samples, which may affect generalizability.

“Many studies lacked external validation and relied on homogeneous population samples, which may affect generalizability.”

May combine two claims.

The improvement in CT-based classification accuracy over time reflects advances in CT imaging resolution, improved three-dimensional reconstruction techniques, standardized cranial landmark identification, and more sophisticated statistical modeling strategies.

“This improvement reflects advances in CT imaging resolution, improved three-dimensional reconstruction techniques, standardized cranial landmark identification, and the adoption of more sophisticated statistical modeling strategies.[,,,,]”

May combine two claims.

Differences in craniofacial sex determination across populations reflect genetic ancestry, environmental conditions, dietary practices, and cultural factors.

“These differences reflect the influence of genetic ancestry, environmental conditions, dietary practices, and cultural factors on craniofacial morphology.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

CBCT offers high spatial resolution for hard tissues with a relatively lower radiation dose.

“CBCT, commonly used in dental and maxillofacial imaging, offers high spatial resolution for hard tissues with a relatively lower radiation dose.”

May combine two claims.

Technological improvements in CT enabled more precise landmark identification and facilitated large-scale morphometric analyses using retrospective clinical CT datasets.

“These technological improvements enabled more precise landmark identification and facilitated large-scale morphometric analyses using retrospective clinical CT datasets.[,,,,]”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The restriction to English-language publications may introduce language bias and potentially exclude relevant regional studies.

“The restriction to English-language publications may introduce language bias and potentially exclude relevant regional studies.”

Classified as probably not a substantial claim.

A structured comparative synthesis was performed, summarizing classification accuracy by geographic region, imaging modality, and statistical approach.

“However, to enhance quantitative interpretability, a structured comparative synthesis was performed. Classification accuracy values were systematically summarized across studies and stratified by geographic region, imaging modality, and statistical approach.”

• AssumptionRests on an assumption: The included studies' reported classification accuracies are sufficiently comparable to permit a structured comparative synthesis despite heterogeneity.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Bizygomatic breadth reflects overall facial breadth and skeletal robusticity, which are typically greater in males due to differences in craniofacial growth and muscle attachment development.

“This measurement reflects overall facial breadth and skeletal robusticity, which are typically greater in males due to differences in craniofacial growth and muscle attachment development.[,,,,,]”

May combine two claims.

Methodological heterogeneity, population-specific variation, and limited external validation remain important limitations of the evidence.

“Overall, the included studies demonstrated a low-to-moderate risk of bias, although methodological heterogeneity, population-specific variation, and limited external validation remain important limitations.”

Probably not a substantial claim: may not pass S2 (checkable).

The results should be interpreted in light of moderate methodological heterogeneity and potential sources of bias.

“The results should be interpreted in light of moderate methodological heterogeneity and potential sources of bias.”

Probably not a substantial claim: may not pass S2 (checkable).

Conventional CT-based craniometric methods retain advantages in transparency, reproducibility, and forensic acceptability, particularly in medicolegal contexts where explainability of results is essential.

“In contrast, conventional CT-based craniometric methods retain advantages in transparency, reproducibility, and forensic acceptability, particularly in medicolegal contexts where explainability of results is essential.”

May combine two claims.

AI-based approaches offer reduced observer dependency and the ability to analyze complex multivariate patterns that may not be captured by traditional statistical methods.

“These approaches offer the advantage of reduced observer dependency and the ability to analyze complex multivariate patterns that may not be captured by traditional statistical methods.”