GAN Based Anomaly Detection in Medical Device and Failure Data
Priya Vijay K, P Ramesh · 2025
In the era of modern health care, medical devices play a vital role henceforth upholding their operational reliability is pivotal for patient welfare. This paper anchors on detecting anomalies in medical device failure data by making use of Generative Adversarial Network (GAN), by signifying dosage related failures such as overdose, underdose and chronic dosing issues. The traditional anomaly detection methods encounter difficulties with high complex and detailed data from medical device data, when peculiarly faced with limited failure example in medical device data. With the deployment of GAN s, the regular functioning of the device can be comprehended along with identifying deviations and fluctuations. By training GAN on a robust dataset of various device failures, anomaly scores for the new data can be generated, to ease the detection of unusual patters in the early stage. This research denotes that with the use of GAN based technology, rather than the customary practice for the detection of anomalies which may fail to be recognized, emphasizing the need to improve device monitoring and reliability. This approach not only increases the precision and reliability of anomaly detection but also, it additionally offers to reveal essential insights of the failure patterns, aiding in the advance maintenance and optimizing risk management techniques. The findings of this research highlight that GANs could contribute to a significant impact, for the detection of anomalies in medical devices, providing a reliable approach for ensuring device safety, effective operations and functionality.