Skip to content. | Skip to navigation

Personal tools
You are here: Home Theses Unsupervised Anomaly Detection in Medical Time Series Data


Prof. Dr. S. Decker
RWTH Aachen
Informatik 5
Ahornstr. 55
D-52056 Aachen
Tel +49/241/8021501
Fax +49/241/8022321

How to find us

Annual Reports





Unsupervised Anomaly Detection in Medical Time Series Data

Thesis type
  • Master
Student Tobias Brockhoff
Status Finished
Submitted in 2019
Proposal on 30. Oct 2018 00:00
Proposal room Seminarraum I5
Add proposal to calendar vCal

Aim of the researh is to investigate recent anomaly detection approaches on possibly multivariate time series data especially from the  medical domain. In the medical context, data from different patients may exhibit different characteristics, therfore the apporach will focus on patient specific anomaly detection. One of the biggest challanges of the doman is there are not enough annatated data sets. In our scenario, the training data neither contains labels nor is cleaned from anomalies. In this thesis, we compare basic approaches with combined methods where combinations are selected with respect to domain requirements.

Document Actions