AI-driven predictive models for determining age and sex from cranial measurements

The ability to accurately determine the age and sex
of individuals from cranial measurements has significant
implications in forensic anthropology and bioarchaeology.
Leveraging artificial intelligence (AI) advancements,
we have developed predictive models that utilize
cranial measurements to estimate these demographic
variables. Additionally, these models have been adapted
to predict cephalometric measurements from nonradiological
data, enhancing their utility in clinical settings.
This study utilized a dataset comprising cranial
measurements from diverse populations. Advanced AI
algorithms, including Support Vector Machines (SVM),
Artificial Neural Networks (ANN), and Deep Neural
Networks (DNN), were employed to develop predictive
models. The dataset was split into training and validation
sets, and cross-validation techniques were applied to
ensure model robustness and prevent overfitting. Feature
selection and hyperparameter tuning were conducted
to optimize model performance. The predictive models
demonstrated high accuracy in estimating sex, with the
SVM and ANN models achieving accuracies between
86% and 91%. The DNN model further improved accuracy,
reaching over 93%. For age prediction, models
treated age as a continuous and categorical variable,
with the constant model achieving a mean absolute error
(MAE) of less than 2 years and the categorical model
achieving an accuracy of approximately 85%. The cephalometric
prediction models showed promising results,
with significant reductions in mean squared error (MSE)
compared to actual clinical measurements. Integrating
AI algorithms in the analysis of cranial measurements
significantly enhances the accuracy of age and sex estimation.
Furthermore, extending these models to predict
cephalometric measurements from non-radiological
data offers a promising tool for clinical applications,
potentially reducing the need for radiographic imaging.
Future research will expand the dataset and refine the
models for broader applicability and improved accuracy.
References
1. Del Bove, A.; Veneziano, A. A Generalised Neural
Network Model to Estimate Sex from Cranial Metric
Traits: A Robust Training and Testing Approach.
Appl. Sci. 2022, 12, 9285.
2. Foreman, A.; Lazenby, R.A. Use of Craniometric
Data in a Generalized Procrustes Analysis to Establish
Human Biological Affinities. Forensic Sci. Int.
1996, 82, 39-53.
3. GADNN: A Revolutionary Hybrid Deep Learning
Neural Network for Age and Sex Determination Utilizing
Cone Beam Computed Tomography Images of
Maxillary and Frontal Sinuses. BMC Med. Res. Methodol.
2023.