The Power of Exclusion using Automated Osteometric Sorting: Pair-Matching

Area of Science:

• Forensic Anthropology
• Bioarchaeology
• Computational Biology

Background:

• Osteometric sorting is crucial for identifying individuals in commingled human remains.
• Existing models require validation and performance testing across diverse datasets.
• Statistical assumptions, such as normality, can impact model accuracy in forensic contexts.

Purpose of the Study:

• To compare the performance of an original osteometric sorting model against two novel models.
• To validate and test the performance of these models using multiple forensic datasets.
• To present a computational approach for dynamic statistical modeling in commingled assemblages.

Main Methods:

• Comparative analysis of three osteometric sorting models: original and two new.
• Validation across three distinct samples: Forensic Data Bank, USS Oklahoma, and DPAA reference.
• Investigation of normality assumptions and application of data transformation; execution of 14,357,220 osteometric t-tests.

Main Results:

• Osteometric sorting models perform reliably even when reference samples deviate from normality.
• The two newly developed models demonstrate superior performance compared to the original model.
• A case study highlights the enhanced exclusion power of the new models in a real-world scenario.

Conclusions:

• The developed osteometric sorting models are robust and effective for identifying human remains.
• One of the new models is recommended to replace the original for future applications.
• Computational methods enhance the ability to analyze complex commingled skeletal assemblages.