CADOES: An interactive machine-learning approach for sex estimation with the pelvis

Coelho, João d’Oliveira; Curate, Francisco

doi:10.1016/j.forsciint.2019.109873

Cited by 37 publications

(12 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The development of software that can be used to estimate sex and calculate posterior probabilities should make the publication of equations and cutoff points obsolete, yet this is a practice that seems to persist. Software for metric sex estimation includes FORDISC (43), mainly for North America, and CADOES (44) and SeuPF (45), which are based on Portuguese populations. As the geographic coverage of such software increases, it will allow for more reliable sex estimations with quantifiable uncertainty, and the publication of dichotomy‐imposing equations should become a thing of the past.…”

Section: Discussionmentioning

confidence: 99%

Getting Rid of Dichotomous Sex Estimations: Why Logistic Regression Should be Preferred Over Discriminant Function Analysis

Bartholdy

Sandoval

Hoogland

et al. 2020

Journal of Forensic Sciences

View full text Add to dashboard Cite

Sex estimation is an important part of creating a biological profile for skeletal remains in forensics. The commonly used methods for developing sex estimation equations are discriminant function analysis (DFA) and logistic regression (LogR). LogR equations provide a probability of the predicted sex, while DFA relies on cutoff points to segregate males and females, resulting in a rigid dichotomization of the sexes. This is problematic because sexual dimorphism exists along a continuum and there can be considerable overlap in trait expression between the sexes. In this study, we used humeral measurements to compare the performance of DFA and LogR and found them to be very similar under multiple conditions. The overall cross‐validated (leave‐one‐out) accuracy of DFA (75.76–95.14%) was slightly higher than LogR (75.76–93.82%) for simple and multiple variable equations, and also performed better under varying sample sizes (94.03% vs. 93.78%). Three of five DFA equations outperformed LogR under the B index, while all five LogR equations outperformed the DFA equations under the Q index. Both methods saw an improvement in overall accuracy (DFA: 86.74–95.79%; LogR: 86.74–95.76%) when individuals with a classification probability lower than 0.80 were excluded. Additionally, we propose a method for calculating additional cutoff points (PMarks) based on posterior probability values. In conclusion, we recommend using LogR over DFA due to the increased flexibility, robusticity, and benefits for future users of the statistical models; however, if DFA is preferred, use of the proposed PMarks facilitates future analysis while avoiding unnecessary dichotomization.

show abstract

Section: Discussionmentioning

confidence: 99%

Getting Rid of Dichotomous Sex Estimations: Why Logistic Regression Should be Preferred Over Discriminant Function Analysis

Bartholdy

Sandoval

Hoogland

et al. 2020

Journal of Forensic Sciences

View full text Add to dashboard Cite

show abstract

“…In binary classification, there are only two classes, which are usually referred to as positive and negative [ 2 , 6 , 14 ]. This allows for two cases of misclassification: false negative (predicting negative when the actual class is positive) and false positive (predicting positive when the actual class is negative), and both are of equal importance when dealing with forensic contexts [ 2 , 13 , 14 ].…”

Section: Methodsmentioning

confidence: 99%

“…Therefore, two methodological approaches are commonly used for sex diagnosis: morphoscopic and osteometric approaches [ 2 ]. The osteometric and morphoscopic features can be successfully analyzed on pelvic and cranial bones [ 5 , 6 , 7 , 8 , 9 ], whereas osteometric methods are commonly most productive using long bones [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ]. One or multiple skeletal elements or combinations of features enable the forensic deduction of sex depending on the degree of fragmentation and presence of a skeletal element [ 4 , 16 , 17 , 18 , 19 , 20 , 21 ].…”

Section: Introductionmentioning

confidence: 99%

Efficiency of the Adjusted Binary Classification (ABC) Approach in Osteometric Sex Estimation: A Comparative Study of Different Linear Machine Learning Algorithms and Training Sample Sizes

Attia

Kholief

Zaghloul

et al. 2022

Biology

View full text Add to dashboard Cite

The adjusted binary classification (ABC) approach was proposed to assure that the binary classification model reaches a particular accuracy level. The present study evaluated the ABC for osteometric sex classification using multiple machine learning (ML) techniques: linear discriminant analysis (LDA), boosted generalized linear model (GLMB), support vector machine (SVM), and logistic regression (LR). We used 13 femoral measurements of 300 individuals from a modern Turkish population sample and split data into two sets: training (n = 240) and testing (n = 60). Then, the five best-performing measurements were selected for training univariate models, while pools of these variables were used for the multivariable models. ML classifier type did not affect the performance of unadjusted models. The accuracy of univariate models was 82–87%, while that of multivariate models was 89–90%. After applying ABC to the crossvalidation set, the accuracy and the positive and negative predictive values for uni- and multivariate models were ≥95%. Sex could be estimated for 28–75% of individuals using univariate models but with an obvious sexing bias, likely caused by different degrees of sexual dimorphism and between-group overlap. However, using multivariate models, we minimized the bias and properly classified 81–87% of individuals. A similar performance was also noted in the testing sample (except for FEB), with accuracies of 96–100%, and a proportion of classified individuals between 30% and 82% in univariate models, and between 90% and 91% in multivariate models. When considering different training sample sizes, we demonstrated that LR was the most sensitive with limited sample sizes (n < 150), while GLMB was the most stable classifier.

show abstract

“…Similarly, the DSP2 software (Diagnose Sexuelle Probabiliste v2) employs LDA for sex estimation based on pelvic measurements derived from populations that represent worldwide variation (Brůžek et al, 2017). A number of free web applications have been developed for sex estimation using the Coimbra Identified Skeletal Collection, such as CADOES that employs pelvic measurements (d'Oliveira Coelho & Curate, 2019), the Ammer‐Coelho application that focuses on the olecranon fossa of the humerus (Ammer et al, 2019), and CalcTalus that estimates sex based on measurements from the talus and calcaneus (Curate et al, 2021). Moreover, the SeuPF software estimates sex based on measurements of the proximal femur, using a reference sample from the Luís Lopes Collection (Portugal) (Curate et al, 2016), while the KKU Sex Estimation online tool focuses on craniometric sex estimation for Thai populations (Techataweewan et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

SexEst: An open access web application for metric skeletal sex estimation

Constantinou

Nikita

2022

Intl J of Osteoarchaeology

View full text Add to dashboard Cite

Skeletal sex estimation is an essential step in any osteoarcheological study; hence, several metric and morphological methods have been developed for this purpose, employing different skeletal elements. This paper has a dual purpose: (1) test the performance of several machine learning classification models for skeletal sex estimation using worldwide samples of cranial and postcranial measurements and (2) present a free web application for the implementation of the models that exhibit the highest accuracy so that the sex of unknown skeletons can be straightforwardly estimated. Regarding the first objective, using the Goldman database of postcranial metrics and the William W. Howells craniometric database, machine learning classification models were constructed for sex prediction. The models were optimized with respect to their hyperparameters and cross‐validated reaching accuracies ranging from 80.8%–89.5% for the postcranial data and 81.2%–87.7% for the cranial data. The models offering the highest rates of correct sex classification (Extreme Gradient Boosting, Light Gradient Boosting, and Linear Discriminant Analysis) were then selected to construct an open access and open source web application, SexEst, for predicting the sex of unknown skeletons.

show abstract

CADOES: An interactive machine-learning approach for sex estimation with the pelvis

Cited by 37 publications

References 53 publications

Getting Rid of Dichotomous Sex Estimations: Why Logistic Regression Should be Preferred Over Discriminant Function Analysis

Getting Rid of Dichotomous Sex Estimations: Why Logistic Regression Should be Preferred Over Discriminant Function Analysis

Efficiency of the Adjusted Binary Classification (ABC) Approach in Osteometric Sex Estimation: A Comparative Study of Different Linear Machine Learning Algorithms and Training Sample Sizes

SexEst: An open access web application for metric skeletal sex estimation

Contact Info

Product

Resources

About