Forensic Profiling – Chemistry and Machine Learning in a Single Fingerprint
In collaboration with the Netherlands Forensic Institute (NFI), we explored how chemical fingerprints can be used to detect gender and drug use using machine learning.
Challenge
Fingerprints have been a key tool in forensic investigations for decades. But what if you don’t just look at the pattern, but also at the chemical composition? Together with the NFI, we examined whether machine learning based on chemical fingerprints can help classify gender and detect drug use.
Our approach
At the Lowlands 2016 festival, we collected both washed and unwashed fingerprints from participants. This chemical data was combined with demographic information from surveys. We developed a system using various machine learning techniques, including neural networks, random forests, and boosted decision trees.
The results were made accessible through a web application, allowing NFI researchers to easily work with the data.
Result
Classifying drug use proved complex due to the variety of substances and high alcohol consumption. Still, the research yielded valuable insights into which chemical elements have predictive value—and which do not.
Our approach demonstrates the potential of AI in forensic research and lays the groundwork for further studies into broader applications of chemical profiling.
Want to learn more?
Curious how AI can support your research or detection challenge? Contact us at info@nolemmings.nl or schedule a meeting with one of our advisors.
