On AI-Based Maintenance of Medical Equipment in Home-Based Settings

The core aim of this proposition is to address the safety of vulnerable people as more and more healthcare is transferred from hospitals to homes. In that context medical equipment also needs to be brought into homes, and where such equipment itself needs to be monitored to ensure the safety of the care recipients. It has been shown that this is a critical issue, and the study of this contribution aims to approach such issues through modern AI-based techniques for predictive maintenance. This paper reflects on the first stage of a project in the context of healthcare taking place in homes, and where medical equipment needs digitized maintenance. Two preliminary experiments are presented, where those serve as prototypes and proof of concepts. On the one hand, ML-models to detect anomalies in data streams were developed and compared, and on the other hand, a radar system was developed where trilateration was used to locate movements of caretakers at home. Here, detecting falls were considered especially high priority, based on falls being common and serious accidents in homes. So far, experiments have shown promising results, and where an LSTM algorithm and an XGBoost algorithm are proposed for anomaly- and fall detections respectively.