New AI tools for rapid identification in forensic investigations are faster than humans at estimating biological sex

CSIRO, a national science institution in Australia, has developed an advanced artificial intelligence (AI) tool to assist in estimating biological sex from human skulls.

AI tools can accelerate accurate skull identification and support investigators when results are needed quickly, such as criminal analysis or severe natural disasters.

Results published in the Science Report show that the AI ​​tool achieves 97% accuracy, significantly outweighing the 82% accuracy achieved by traditional methods used by human evaluators.

This tool was developed in collaboration with the University of Western Australia (UWA). That forensic anthropology expert provided labeled data and domain knowledge to support model development.

CSIRO research scientist and research co-scientist, Dr. Hollie Min, has explained the characteristics of sexual relevance (CT scans) from 200 computerized tomography or CT scan datasets. He said he analyzed results and analyzed results compared to human analysis. .

“Our AI tools produce results about five times faster than human capabilities, meaning families awaiting the results of their investigations can receive news about their loved ones more quickly. ” said Dr. Min. “This AI tool has the potential to help forensic anthropologists improve the accuracy of sexual estimation and reduce the potential impact of human bias.”

Dr. Min also emphasized the importance of considering population-specific variation in skull characteristics.

“This collaboration has been able to address some of the perceived limitations of traditional methods and better explain the diversity of forensic data,” she said. “Future research is needed, in particular, by expanding the dataset to include diverse populations, it will improve the robustness and generalizability of the AI ​​framework.

“Our goal is to support their important work with reliable and interpretable tools for forensic anthropologists, especially when individuals from unknown population backgrounds are involved.”

This collaboration demonstrates the potential for AI to support forensic anthropology and advance the field with innovative, data-driven solutions.

“Our team is currently looking for industry collaborators to develop and translate this technology for real-world applications,” added Dr. Min.

The CT database was collected at Wahidin Sudirohusodo General Hospital (RSW) at Hasanuddin University, Indonesia.