Original title: Sex estimation from anteroposterior knee radiographs in a Brazilian sample
Researchers measured three distances on X-rays of the left knee from adults in Brazil. They used each measurement on its own to build a model that estimates whether a person was female or male. The best-performing measurement, TPML, reached 81% accuracy when the researchers repeatedly tested the model on different parts of their sample. The authors say knee X-rays could help forensic investigations, but the models need more testing.
Imagine sorting a pile of keys by measuring their lengths. Some keys from different groups will still be almost the same size, so one measurement cannot sort every key correctly. In this paper, knee measurements are the key lengths, and the sorting is an estimate of whether a person was female or male. Testing the rule on different portions of the pile checks how often it sorts correctly within that collection.
When human remains are difficult to identify, existing knee radiographs may offer another clue. The results suggest these measurements could support forensic investigations, but the models still need testing in independent populations before their reliability elsewhere is clear.
The models were tested by cross-validation within the study sample, not in independent populations. The authors say further studies should use calibrated images and external validation to strengthen the models' robustness, generalizability, and forensic applicability.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Sex estimation from anteroposterior knee radiographs in a Brazilian sample — Mirela Cristina da Silva, I. Moreira, Paulo Henrique Viana Pinto et al. — Egyptian Journal of Forensic Sciences — doi:10.1186/s41935-026-00585-y
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.