Segmented Bayesian calibration approach for estimating age in forensic science.

Area of Science:

• Forensic Science
• Biostatistics
• Radiology

Background:

• Forensic age estimation is crucial for identifying individuals and assessing human remains.
• Current radiological methods often use regression models, which can introduce age-related biases.
• Existing models tend to overestimate ages in younger subjects and underestimate them in older individuals.

Purpose of the Study:

• To develop a novel Bayesian calibration method combined with a segmented function for more accurate forensic age estimation.
• To mitigate the inherent biases of traditional regression models in age assessment.
• To effectively model decreasing growth rates in juvenile populations.

Main Methods:

• A full Bayesian calibration method was employed.
• A segmented function was integrated with a Normal distribution density model.
• The new model was compared against existing age estimation approaches.
• The method was validated on a dataset of South African juvenile subjects.

Main Results:

• The proposed Bayesian-segmented model demonstrated more robust and precise age forecasts compared to existing methods.
• The accuracy of the new method was comparable to established forecasting measures.
• The model successfully mitigated estimation bias, particularly in younger subjects.
• Validation on real-world data confirmed the method's effectiveness.

Conclusions:

• The developed Bayesian-segmented model offers a significant improvement in forensic age estimation accuracy.
• This method effectively addresses the overestimation and underestimation biases present in traditional regression models.
• The approach provides a more reliable tool for age assessment in both living individuals and human remains, especially for juvenile populations.