Using posterior probability informed thresholds to develop best practice recommendations for MorphoPASSE using the...

Area of Science:

• Forensic anthropology
• Bioinformatics
• Skeletal biology

Background:

• MorphoPASSE is a free program for sex estimation using morphological traits.
• Current limitations include lack of guidance on optimal trait sets and posterior probability thresholds.
• Standardized reporting protocols are needed for forensic casework.

Purpose of the Study:

• Compare accuracy rates of MorphoPASSE using different trait combinations (innominate, cranium, combined).
• Evaluate the effectiveness of posterior probability informed thresholds (PPITs) for sex classification.
• Provide recommendations for optimizing MorphoPASSE use in modern forensic anthropology.

Main Methods:

• Collected innominate and cranial trait data from 285 individuals in US skeletal collections.
• Calculated accuracy rates for five posterior probability (PP) intervals.
• Condensed significant PP intervals into PPITs for analysis.
• Compared accuracy between innominate, cranial, and combined trait models.

Main Results:

• A PPIT of 0.85-1.00 yielded high accuracy: 97.3% for innominate and 90.2% for cranial traits.
• Combined innominate and cranial traits achieved 99.6% accuracy with a lower PPIT of 0.75-1.00.
• The combined model corrected previous misclassifications, either classifying sex or leaving individuals unclassified.

Conclusions:

• The combined set of innominate and cranial traits is recommended for sex estimation using MorphoPASSE when both are available.
• A PPIT of 0.75-1.00 is proposed for reporting sex estimation results.
• Standardized reporting using these recommendations will enhance consistency in forensic casework.