Original title: Machine learning-based comparative analysis of nasal morphology: Forensic implications for sexual dimorphism and ethnic variation
Researchers measured eighteen visible nose traits in photos of 500 adults from two North Indian groups. They trained three computer methods to predict each person's sex or group from those traits. The best method was correct 69% of the time for sex and 75% for group. Nose-root height and nose size were steady clues for sex, while group predictions depended on more varied traits.
Imagine sorting a pile of mixed socks by who they belong to. Some pairs have clear clues, while others look much alike. In this paper, the visible nose traits are the clues, and the sorting task is predicting a person's sex or which of the two groups they belong to.
Forensic teams sometimes use facial features to help estimate a person's identity. This study suggests nasal traits may be more useful for distinguishing the two groups it examined than for estimating sex, but the predictions are not certain enough to stand alone.
The abstract covers adults aged 18–35 from only two North Indian populations, so it does not establish how well these results apply to other ages, populations, or settings. The best reported accuracy is 75% for ethnicity and 69% for sex; the abstract does not describe external testing or establish that nasal traits can identify an individual.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Machine learning-based comparative analysis of nasal morphology: Forensic implications for sexual dimorphism and ethnic variation — Tej Kaur, Damini Siwan, Ankita Guleria et al. — Egyptian Journal of Forensic Sciences — doi:10.1186/s41935-026-00589-8
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.