Criminal investigation: a labyrinth in which every detail can be a key to the truth. How do analysts navigate through huge amounts of data to link individual crimes and uncover the activities of serial criminals, for example? Today, artificial intelligence (AI) is increasingly being used to support such tasks.
Jessica Woodhams, Professor of Forensic Psychology at the University of Birmingham, who has been researching crime linking for over 20 years, explains: “Linking crimes is a complex task that requires considerable effort and insight on the part of analysts.“
Prof. Dr. J. Woodhams actively collaborates with police authorities in the United Kingdom and other countries to ensure that research and interventions have a real impact on police practice. She has received several awards for the practical application of psychological research. She also co-edited the first book on crime linkage with Prof. Dr. Craig Bennell and founded the International Crime Linkage Network (C-LINK) together with Prof. Dr. Matt Tonkin and Dr. Amy Burrell. The network has been operating successfully for over 10 years.
Last week, she gave a presentation in Vilnius as a guest speaker at the annual conference of the European Association for Psychology and Law (EAPL), EAPL 2025, which was attended by around 250 participants from 34 countries. The conference was organised by the Laboratory for Applied Psychology Research at Mykolas Romeris University (MRU).
As a guest speaker at EAPL 2025, the professor presented LATIS – a decision support tool developed for the UK National Crime Agency (NCA). This tool, which was developed in collaboration with several universities and the NCA, analyses behaviour at crime scenes to help analysts identify related sexual offences.
How does it work?
LATIS uses algorithms that analyse various aspects of a crime, including geographical location, time and behavioural patterns. Based on this information, the tool can generate probabilities that provide insight into which crimes are most likely to be related.
“Our algorithm uses information that has been proven to be the most accurate in predicting whether multiple crimes are related,“ explains Prof. Dr. J. Woodhams. “This information is presented to the analyst so that they can make an empirically sound decision.“
Inspiration and development process
The idea for developing this tool arose from the personal experiences of Prof. Dr. J. Woodhams and her many years of working with criminal analysts in the field.
“Over 20 years ago, I worked as a criminal analyst and experienced first-hand how difficult it was to link crimes together,’ she recalls. ‘Working with analysts, we found that this task was becoming increasingly complex as their databases continued to grow.“
During the development of LATIS, great importance was attached to practicality and usefulness. “From the outset, we worked closely with analysts to understand where the tool would be most valuable for their work,“ explains Prof. Dr. J. Woodhams. “The Serious Crime Analysis Unit (the NCA's analysis unit) was involved throughout the research and software development process, helping to design the user interfaces and evaluating the tool with us.“
Future prospects
Prof. Dr. J. Woodhams is convinced that LATIS has the potential to significantly change police work and criminal investigations in the future. She hopes that the tool will speed up the process of identifying links between crimes and help analysts make more efficient decisions.
“I hope that this will enable units such as the Serious Crime Analysis Unit to analyse a larger number of cases more quickly,“ she says. “This means that their analysis results will reach investigators faster, which could offer an additional advantage – new leads could emerge at an earlier stage of the investigation.“
Ethical Aspects
When developing AI tools for criminal justice, it is essential to take ethical and data privacy concerns into account. Prof. Dr. J. Woodhams emphasizes that protecting the identities of both victims and offenders was a priority in the research. All information is anonymized, and strict protocols are applied for data storage and access.
“It is also important to consider whether the data used to develop the tool might be biased," she adds. "That’s why we implemented a comprehensive research program to check the data for possible biases, and so far, we have found no evidence that the data or the tool’s outputs are biased.“
In the future, Prof. Dr. J. Woodhams and her colleagues plan to further refine and expand this research, collaborate with similar institutions worldwide, and develop decision-making tools for other types of crime.
This AI tool represents a step forward in the fight against crime, providing analysts with a powerful instrument to uncover the activities of serial offenders and ensure justice.