Siam Knecht
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Siam Knecht has a total of 30 co-authors. Here are the top co-authors by number of shared publications:
Christophe Roman
LIS UMR 7020 CNRS, Aix Marseille Université, Marseille, France
4 shared publicationsCristina Cattaneo
LABANOF (Laboratorio di Antropologia e Odontologia Forense), Department of Biomedical Science for Health University of Milan Milan Italy
3 shared publicationsMustapha Ouladsine
LIS UMR 7020 CNRS, Aix Marseille Université, Marseille, France
3 shared publicationsPascal Adalian
Aix Marseille Univ, CNRS, EFS, ADES, Marseille, France
3 shared publicationsGabriele Krüger
University of Pretoria, Pretoria, South Africa
2 shared publicationsLeandi Liebenberg
University of Pretoria, Pretoria, South Africa
2 shared publications
Latest Publications
- Open Access
- 13 October 2026
- Published by Wiley in American Journal of Biological Anthropology
- Vol. 191 (2)
Constancy of Sexual Dimorphism in the Infracranial Skeleton
Sexual dimorphism in the human infracranial skeleton has been extensively studied, yet its structural organization across skeletal regions remains incompletely characterized at a global scale. Using a geographically diverse sample of 3177 individuals of known sex (1219 females; 1958 males) drawn from five documented skeletal collections, this study quantifies sexual dimorphism across 12 long bones measurements of the humerus, radius, femur, and tibia. Dimorphism indices (DS%), two‐way ANOVAs with partial η 2 , Mahalanobis distances, Hedges' g , and D ‐index were computed to characterize the magnitude and the consistency of sex differences across groups. Results reveal a clear anatomical gradient in the expression of sexual dimorphism. Maximum bone lengths exhibit the lowest dimorphism (DS% = 8.5%–11.5%) with high stability and low within‐sex variability. Articular breadths show the strongest association with sex (partial η 2 > 56%) combined with relatively low intra‐sexual dispersion. Diaphyseal diameters, particularly in the upper limb, display the highest DS% values but also the greatest coefficients of variation, reflecting greater biomechanical and methodological sensitivity. Sex × Group interactions are weak or non‐significant across all variables, and rank concordance analyses confirm that this hierarchical pattern is consistent across populations (Kendall's W = 0.707–0.814, p < 0.001). These findings strongly support previous studies demonstrating the reliability of articular breadths for sex estimation in skeletal remains, and confirm that the metric signal of sexual dimorphism in long bones reflects a fundamental and broadly shared structural organization of the human appendicular skeleton.
- 13 October 2026
- Published by Wiley in American Journal of Biological Anthropology
- Vol. 191 (2)
Open AccessConstancy of Sexual Dimorphism in the Infracranial Skeleton
Sexual dimorphism in the human infracranial skeleton has been extensively studied, yet its structural organization across skeletal regions remains incompletely characterized at a global scale. Using a geographically diverse sample of 3177 individuals of known sex (1219 females; 1958 males) drawn from five documented skeletal collections, this study quantifies sexual dimorphism across 12 long bones measurements of the humerus, radius, femur, and tibia. Dimorphism indices (DS%), two‐way ANOVAs with partial η 2 , Mahalanobis distances, Hedges' g , and D ‐index were computed to characterize the magnitude and the consistency of sex differences across groups. Results reveal a clear anatomical gradient in the expression of sexual dimorphism. Maximum bone lengths exhibit the lowest dimorphism (DS% = 8.5%–11.5%) with high stability and low within‐sex variability. Articular breadths show the strongest association with sex (partial η 2 > 56%) combined with relatively low intra‐sexual dispersion. Diaphyseal diameters, particularly in the upper limb, display the highest DS% values but also the greatest coefficients of variation, reflecting greater biomechanical and methodological sensitivity. Sex × Group interactions are weak or non‐significant across all variables, and rank concordance analyses confirm that this hierarchical pattern is consistent across populations (Kendall's W = 0.707–0.814, p < 0.001). These findings strongly support previous studies demonstrating the reliability of articular breadths for sex estimation in skeletal remains, and confirm that the metric signal of sexual dimorphism in long bones reflects a fundamental and broadly shared structural organization of the human appendicular skeleton.
- Open Access
- 8 September 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (18) , 2893
Machine Learning-Based Sex Estimation from Computed Tomography-Derived Pelvic Morphometric Measurements in a Contemporary Turkish Population
Background/Objectives: Sex estimation is a fundamental component of the forensic biological profile, and the pelvis is widely regarded as the most sexually dimorphic skeletal element. This study aimed to quantify pelvic sexual dimorphism in a contemporary Turkish sample using computed tomography (CT)-derived morphometric measurements, and to evaluate the performance of several machine learning (ML) algorithms for sex classification. Methods: Fourteen pelvic measurements were obtained from CT reconstructions of 201 individuals (101 males, 100 females) by a single observer; a stratified subset of 30 cases (15 males, 15 females) was remeasured by the same observer and by a second observer to assess intra- and inter-observer reliability. Four supervised ML classifiers—logistic regression, linear discriminant analysis (LDA), random forest, and support vector machine (SVM)—were trained and evaluated using stratified 10-fold cross-validation and an independent 30% hold-out test set. Results: Thirteen of the fourteen measurements differed significantly between sexes (p < 0.05). Intra-observer reliability was excellent for all 14 measurements (ICC = 0.943–0.998), and inter-observer reliability was good to excellent (ICC = 0.834–0.996), with the lowest value observed for pubic length. Cross-validated classification accuracy ranged from 98.0% (random forest) to 99.0% (logistic regression, LDA, and SVM), with corresponding cross-validated AUC values of 0.991–0.994, with logistic regression, LDA, and SVM each achieving 100% accuracy (AUC = 1.000) on the independent hold-out set (random forest: 96.7%, AUC = 0.999). Acetabular width was identified as the single most informative predictor (univariate area under the curve = 0.972), followed by acetabular height, the subpubic angle, ischial length, the transverse pelvic outlet, and the angle of the greater sciatic notch. Conclusions: These findings demonstrate that CT-derived pelvic morphometrics, combined with machine learning classification, provide a highly accurate and reproducible method for sex estimation in a Turkish forensic context, and support the development of population-specific standards for forensic anthropological casework, pending independent external validation; the classification and effect-size estimates reported here were, however, stable under repeated cross-validation and hyperparameter optimisation.
- 8 September 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (18) , 2893
Open AccessMachine Learning-Based Sex Estimation from Computed Tomography-Derived Pelvic Morphometric Measurements in a Contemporary Turkish Population
Background/Objectives: Sex estimation is a fundamental component of the forensic biological profile, and the pelvis is widely regarded as the most sexually dimorphic skeletal element. This study aimed to quantify pelvic sexual dimorphism in a contemporary Turkish sample using computed tomography (CT)-derived morphometric measurements, and to evaluate the performance of several machine learning (ML) algorithms for sex classification. Methods: Fourteen pelvic measurements were obtained from CT reconstructions of 201 individuals (101 males, 100 females) by a single observer; a stratified subset of 30 cases (15 males, 15 females) was remeasured by the same observer and by a second observer to assess intra- and inter-observer reliability. Four supervised ML classifiers—logistic regression, linear discriminant analysis (LDA), random forest, and support vector machine (SVM)—were trained and evaluated using stratified 10-fold cross-validation and an independent 30% hold-out test set. Results: Thirteen of the fourteen measurements differed significantly between sexes (p < 0.05). Intra-observer reliability was excellent for all 14 measurements (ICC = 0.943–0.998), and inter-observer reliability was good to excellent (ICC = 0.834–0.996), with the lowest value observed for pubic length. Cross-validated classification accuracy ranged from 98.0% (random forest) to 99.0% (logistic regression, LDA, and SVM), with corresponding cross-validated AUC values of 0.991–0.994, with logistic regression, LDA, and SVM each achieving 100% accuracy (AUC = 1.000) on the independent hold-out set (random forest: 96.7%, AUC = 0.999). Acetabular width was identified as the single most informative predictor (univariate area under the curve = 0.972), followed by acetabular height, the subpubic angle, ischial length, the transverse pelvic outlet, and the angle of the greater sciatic notch. Conclusions: These findings demonstrate that CT-derived pelvic morphometrics, combined with machine learning classification, provide a highly accurate and reproducible method for sex estimation in a Turkish forensic context, and support the development of population-specific standards for forensic anthropological casework, pending independent external validation; the classification and effect-size estimates reported here were, however, stable under repeated cross-validation and hyperparameter optimisation.
- 20 June 2026
- Published by Springer Nature in International journal of legal medicine
- Vol. 140 (6) , 4027-4036
A robust forensic reasoning to estimate the minimum number of individuals: the certain exclusion
Estimating the Minimum Number of Individuals (MNI) in commingled skeletal assemblages is a core task in forensic and archaeological anthropology. Traditional methods, based on the frequency of lateralized elements, often suffer from underestimation due to fragmentation and morphological similarity. This study introduces a new MNI estimation method grounded in the principle of certain exclusion (with a 99% confidence interval), which focuses on detecting osteometric incompatibilities between elements rather than identifying matches. We developed linear regression models based on a large international dataset (n = 2,969 individuals) to generate 99% prediction intervals between long bone measurements. A recursive protocol, termed Allometric Research by Exclusion (RAE), was implemented to isolate bones that cannot belong to the same individual. The method was validated via bootstrap simulations using independent samples from the Milano and Pretoria Bone collections. Across all assemblages, the exclusion-based MNI estimates were consistently higher than those obtained by frequency-based methods, and more closely approximated the real number of individuals. This trend was especially clear in small to mid-sized assemblages, where precise MNI estimation is most critical. The exclusion method offers a statistically grounded, replicable, and conservative approach to MNI estimation. Its integration into forensic and archaeological workflows may enhance the accuracy and defensibility of population reconstruction in complex contexts.
- 20 June 2026
- Published by Springer Nature in International journal of legal medicine
- Vol. 140 (6) , 4027-4036
A robust forensic reasoning to estimate the minimum number of individuals: the certain exclusion
Estimating the Minimum Number of Individuals (MNI) in commingled skeletal assemblages is a core task in forensic and archaeological anthropology. Traditional methods, based on the frequency of lateralized elements, often suffer from underestimation due to fragmentation and morphological similarity. This study introduces a new MNI estimation method grounded in the principle of certain exclusion (with a 99% confidence interval), which focuses on detecting osteometric incompatibilities between elements rather than identifying matches. We developed linear regression models based on a large international dataset (n = 2,969 individuals) to generate 99% prediction intervals between long bone measurements. A recursive protocol, termed Allometric Research by Exclusion (RAE), was implemented to isolate bones that cannot belong to the same individual. The method was validated via bootstrap simulations using independent samples from the Milano and Pretoria Bone collections. Across all assemblages, the exclusion-based MNI estimates were consistently higher than those obtained by frequency-based methods, and more closely approximated the real number of individuals. This trend was especially clear in small to mid-sized assemblages, where precise MNI estimation is most critical. The exclusion method offers a statistically grounded, replicable, and conservative approach to MNI estimation. Its integration into forensic and archaeological workflows may enhance the accuracy and defensibility of population reconstruction in complex contexts.
- Open Access
- 28 February 2026
- Published by Wiley in Journal of Forensic Sciences
- Vol. 71 (3) , 1388-1397
Population‐specific validation of long bone linear discriminant analysis against morphoscopic pelvic methods for sex estimation in contemporary Italian skeletal remains
Sex estimation represents a fundamental step in forensic identification protocols, traditionally relying on morphoscopic pelvic assessment. However, the increasing integration of machine learning approaches and population‐specific validation requirements necessitate comprehensive evaluation of alternative methodologies. This study provides the first direct comparison of established morphoscopic methods (Phenice 1969, Bruzek 2002) against a multivariable long bone linear discriminant analysis (LDA) model in a sample of 333 documented individuals from the CAL Milano Cemetery Skeletal Collection. Metric data were preprocessed to address missing values through imputation prior to analysis. The morphoscopic methods achieved high accuracy rates (Phenice: 96.9%, Bruzek: 97.7%) but showed significant exclusion rates due to preservation limitations (32.4% and 2.4% respectively). The long bone LDA model demonstrated comparable performance with threshold‐dependent accuracy ranging from 95.2% (0.50 threshold) to 98.1% (0.88 threshold), while maintaining universal applicability across all specimens. Crucially, disagreement analysis revealed method‐specific error patterns with minimal overlap in misclassified individuals, supporting complementary rather than redundant diagnostic signals. These findings validate long bones as a reliable alternative for sex estimation in fragmentary remains while establishing population‐specific accuracy benchmarks for contemporary Italian forensic applications. The threshold‐adjustable probabilistic framework offers operational flexibility for balancing classification certainty against sample coverage requirements.
- 28 February 2026
- Published by Wiley in Journal of Forensic Sciences
- Vol. 71 (3) , 1388-1397
Open AccessPopulation‐specific validation of long bone linear discriminant analysis against morphoscopic pelvic methods for sex estimation in contemporary Italian skeletal remains
Sex estimation represents a fundamental step in forensic identification protocols, traditionally relying on morphoscopic pelvic assessment. However, the increasing integration of machine learning approaches and population‐specific validation requirements necessitate comprehensive evaluation of alternative methodologies. This study provides the first direct comparison of established morphoscopic methods (Phenice 1969, Bruzek 2002) against a multivariable long bone linear discriminant analysis (LDA) model in a sample of 333 documented individuals from the CAL Milano Cemetery Skeletal Collection. Metric data were preprocessed to address missing values through imputation prior to analysis. The morphoscopic methods achieved high accuracy rates (Phenice: 96.9%, Bruzek: 97.7%) but showed significant exclusion rates due to preservation limitations (32.4% and 2.4% respectively). The long bone LDA model demonstrated comparable performance with threshold‐dependent accuracy ranging from 95.2% (0.50 threshold) to 98.1% (0.88 threshold), while maintaining universal applicability across all specimens. Crucially, disagreement analysis revealed method‐specific error patterns with minimal overlap in misclassified individuals, supporting complementary rather than redundant diagnostic signals. These findings validate long bones as a reliable alternative for sex estimation in fragmentary remains while establishing population‐specific accuracy benchmarks for contemporary Italian forensic applications. The threshold‐adjustable probabilistic framework offers operational flexibility for balancing classification certainty against sample coverage requirements.
- Top Cited PapersOpen Access
- 10 January 2026
- Published by Elsevier in Forensic Science International
- Vol. 380, 112820
Case-specific accuracy in sex estimation from long bones in forensic anthropology: An “accuracy x-factors” approach
- 10 January 2026
- Published by Elsevier in Forensic Science International
- Vol. 380, 112820
Top Cited PapersOpen AccessCase-specific accuracy in sex estimation from long bones in forensic anthropology: An “accuracy x-factors” approach
- 25 October 2025
- Published by Springer Nature in International journal of legal medicine
- Vol. 140 (2) , 983-995
Interpretable machine learning for individualized sex estimation from long bones
Sex estimation is an essential task in forensic anthropology. It is not only crucial for the identification of individuals from skeletal remains, but it is also essential for improving the reliability of other methods of biological profile estimation, such as age and stature, some of which perform better when sex is taken into account. This study investigates the application of machine learning (ML) techniques to sex estimation, with a particular focus on interpretability to address the “black box” challenge inherent in AI models. Using a diverse dataset of long bone measurements from 2,969 individuals, 12 different ML algorithms were evaluated. Missing data were handled using iterative regression imputation, though challenges arising from incomplete datasets underscored the need for improved data handling strategies. Linear Discriminant Analysis (LDA) emerged as the most accurate approach, achieving 95.2% accuracy. A key feature of this study is the integration of SHapley Additive exPlanations (SHAP) values, which provide individualized insights into the factors influencing each prediction. This interpretability framework ensures transparency and addresses legal and scientific concerns about the admissibility of AI-generated evidence in court. Indeed, misclassifications possibilities highlight the importance of clear, understandable models in forensic applications. The study emphasizes the significance of individualized prediction, illustrated by the probability of male or female classification for each individual, as well as the impact of missing values on prediction accuracy. This research demonstrates that ML models can effectively balance accuracy with interpretability, offering personalized, actionable insights for forensic investigations. It paves the way for AI-driven methods that meet both scientific rigor and legal standards, transforming sex estimation in forensic science by providing individualized, defensible evidence suitable for court.
- 25 October 2025
- Published by Springer Nature in International journal of legal medicine
- Vol. 140 (2) , 983-995
Interpretable machine learning for individualized sex estimation from long bones
Sex estimation is an essential task in forensic anthropology. It is not only crucial for the identification of individuals from skeletal remains, but it is also essential for improving the reliability of other methods of biological profile estimation, such as age and stature, some of which perform better when sex is taken into account. This study investigates the application of machine learning (ML) techniques to sex estimation, with a particular focus on interpretability to address the “black box” challenge inherent in AI models. Using a diverse dataset of long bone measurements from 2,969 individuals, 12 different ML algorithms were evaluated. Missing data were handled using iterative regression imputation, though challenges arising from incomplete datasets underscored the need for improved data handling strategies. Linear Discriminant Analysis (LDA) emerged as the most accurate approach, achieving 95.2% accuracy. A key feature of this study is the integration of SHapley Additive exPlanations (SHAP) values, which provide individualized insights into the factors influencing each prediction. This interpretability framework ensures transparency and addresses legal and scientific concerns about the admissibility of AI-generated evidence in court. Indeed, misclassifications possibilities highlight the importance of clear, understandable models in forensic applications. The study emphasizes the significance of individualized prediction, illustrated by the probability of male or female classification for each individual, as well as the impact of missing values on prediction accuracy. This research demonstrates that ML models can effectively balance accuracy with interpretability, offering personalized, actionable insights for forensic investigations. It paves the way for AI-driven methods that meet both scientific rigor and legal standards, transforming sex estimation in forensic science by providing individualized, defensible evidence suitable for court.