MARVIN aims to enhance anti-doping measures by developing an advanced AI-based tool for detecting identical steroid profiles in athletes. The project addresses a specific doping method where athletes swap their urine samples with either a substitute's sample or their own sample taken before the use of the prohibited substance. This method, though previously detected, poses challenges in identification, especially when samples are processed at different times or locations.
Building on the groundwork laid by the 3S Project, which utilized machine learning to identify atypical profiles resulting from sample swapping, MARVIN shifts focus to the detection of identical steroid profiles from the same athlete. The key objective is to create a screening tool leveraging artificial intelligence to analyze the extensive data from the Athlete Biological Passport (ABP) steroidal module.
The core objective is to develop a pattern recognition and classification system that compares and scores the similarity of steroid profiles. This involves thorough research into athletes' steroid profiles, identifying key indicators using statistical methods, implementing cutting-edge deep learning technologies, and constructing a comprehensive pipeline for detecting fraudulent samples. Project MARVIN represents a significant step forward in the fight against doping in sports, leveraging advanced technology for fair and clean competition.
Partners
Deutsche Sporthochschule Köln (DSHS Köln)
