The Automation of Regression Modeling in Osteometric Sorting: An Ordination Approach.

Area of Science:

• Forensic Anthropology
• Bioarchaeology
• Human Osteology

Background:

• Osteometric sorting is crucial for identifying commingled human remains.
• Current methods rely on traditional statistical approaches.
• Advancements are needed for improved accuracy in forensic casework.

Purpose of the Study:

• To compare the efficacy of the original osteometric sorting method with a novel ordination approach.
• To evaluate performance across various skeletal element combinations and measurement types.
• To determine the optimal method for large-scale human remains identification.

Main Methods:

• Utilized the Defense POW/MIA Accounting Agency osteometric reference sample.
• Applied both original (prediction interval, t-statistic) and ordination methods.
• Tested models with and without standard length measurements for humerus, ulna, radius, femur, tibia, and fibula.

Main Results:

• The ordination approach demonstrated superior performance over the original method in most bone combinations.
• Models incorporating length measurements exhibited greater exclusion power.
• The ordination method showed higher accuracy and efficiency in skeletal element association.

Conclusions:

• The ordination approach is recommended to replace the original method for identifying commingled human remains.
• Incorporating length measurements enhances the discriminatory power of osteometric sorting.
• This refined method offers improved accuracy for forensic anthropological analyses.