Publication
Incorporating Metabolic Information into LLMs for Anomaly Detection in Clinical Time-Series
Maxx Richard Rahman; Ruoxuan Liu; Wolfgang Maaß
In: Time Series in the Age of Large Models Workshop at Neural Information Processing Systems (NeurIPS). Neural Information Processing Systems (NeurIPS-2024), NeurIPS, 2024.
Abstract
Anomaly detection in clinical time-series holds significant potential in identifying suspicious patterns in different biological parameters. This paper proposes a targeted method that incorporates the clinical domain knowledge into LLMs to improve their ability to detect anomalies. The Metabolism Pathway-driven Prompting (MPP) approach is introduced, which integrates the information about metabolic pathways to better capture the structural and temporal changes in biological samples. We applied our method for doping detection in sports, focusing on steroid metabolism, and evaluated using real-world data from athletes. The results show that our method improves anomaly detection performance by leveraging metabolic context, providing an improved prediction of suspicious samples in athletes’ profiles.
Projects
- 3S Project - Steroid Sample Swapping
- EPOPredictII - Erythropoietin Prediction II
- MARVIN - Development of AI-based Screening Tool to detect Identical Urine Samples within the Athlete Biological Passport
- 3S-II - AI-based detection of steroid sample swapping in the anti-doping area and analysis of the associated decision-making processes