Testing the population affinity and revising the ‘JSDNE’ R package with Southeast Asian population

Age estimation is a key component in constructing the biological profile of the deceased, playing a vital role in forensic anthropology and bioarchaeology. Jang et al. (2024b) developed the JSDNE R package, incorporating three models: PCQDA, PCR, and PCLR. This method quantitatively assesses curvature on reconstructed auricular surfaces using Dirichlet Normal Energy (DNE). All three models demonstrated high accuracy across age cohorts. However, the JSDNE package was developed using modern European skeletal samples, raising concerns about statistical bias and broad applicability to diverse populations. This study evaluates the population affinity of the JSDNE package with the Southeast Asian sample and compares degenerative patterns across populations. A total of 335 adult auricular surfaces in the Khon Kaen University (KKU) Human Skeletal Research Centre (HSRC), Thailand, were reconstructed into 3D models. The PCQDA, PCR, and PCLR models produced accuracy levels of 64%, 35%, and 66%, respectively. Due to different degenerative patterns and a narrow age range, the PCR model displayed distinct population affinity. To address this, the JSDNE package was refined by developing the A_PCR model with Southeast Asian samples and the M_PCR model with multi-populational datasets. The A_PCR and the M_PCR models achieved 69% and 67% accuracy for the Southeast Asian sample. The M_PCR model provides 68% accuracy for the European sample. The detailed code of this study is openly available at https://github.com/jisunjang19/cran-JSDNE or doi: 10.5281/zenodo.12708778.