Abstract: Anomaly detection techniques are well-known in machine learning. Tools and methods like dynamically moving average, isolation forests or thresholds are used often. Although they are defined for general use, they often inhibit different results if used in computer-generated data or human-generated data. In this seminar, I will present the latest results from the associated project microHRV, where we use anomaly detection techniques to help critically ill patients (e.g. stroke) and to recognize events in radio networks (e.g. rain or wind). I will discuss the methods used, their limitations and show how to find the best methods in both cases. You will learn what the anomaly detection techniques show, how to interpret their results and when to trust them.