Sex Estimation from Fragmented Thai Femora: Developing Segment-Specific Models Using Discriminant Function Analysis

Background: Sex estimation from skeletal remains is important for forensic identification, but many methodologies focus on complete elements despite high fragmentation rates in operational contexts. The aim of this study was to develop and validate discriminant function equations for sex estimation between complete and fragmented Thai femora. Materials and Methods: A total of 560 adult femora (280 males and 280 females) were used for measurements of eight osteometric variables. Then, discriminant function analysis was applied to complete femora and anatomically isolated segments, including proximal, diaphyseal, and distal, with leave-one-out cross-validation. Results: All measurements showed significant sexual dimorphism, with percentage differences ranging from 6.56% to 42.27%. Complete femur stepwise analysis achieved 90.47% accuracy by using four optimally selected variables, performing comparably to eight-variable models. Isolated segment accuracies varied substantially: proximal segments achieved 89.64% accuracy, differing by only 0.83 percentage points from complete performance; distal segments demonstrated 86.25% accuracy from bicondylar width alone; and diaphyseal segments achieved 80.88%. Combined segment approaches demonstrated synergistic effects approaching complete femur performance. Conclusions: These population-specific equations provided validated methodologies for sex estimation from complete and fragmentary Thai femora. Anatomical region selection still maintained high classification accuracy despite skeletal incompleteness, in line with the fragmentary conditions commonly encountered in forensic and archeological contexts.