Enhancing patient safety through integrated sensor technology and machine learning for bed-based patient movement detection in inpatient care

Abstract

The occurrence of inpatient falls and new-onset seizures are common complications during hospital stays, posing risks to patient safety and potentially leading to prolonged hospital stays and further complications. Given the constraints on medical staff’s ability to provide constant monitoring due to their workload, the implementation of a sensor device equipped with machine learning capabilities to recognize and prevent these events becomes imperative. This study utilized data acquired through the Movella Xsens sensor, which detects real-time motions and 3D movements, in conjunction with the PyCaret machine-learning algorithm. Adult-sized and infant-sized mannequins were used to assess the algorithm’s ability in predicting specific movements associated with breathing, seizures, rolling to the right side, rolling to the left side, rolling off the bed from the left, and rolling off the bed from the right. The study achieved an overall 89% accuracy rate in detecting each specific movement using the combination of PyCaret and Xsens sensors. The application of PyCaret alongside Xsens sensors demonstrates promising results in accurately detecting movements, thereby mitigating falls and post-seizure complications in an inpatient setting, consequently improving patient safety. Further exploration of this technology holds the potential to revolutionize healthcare delivery by incorporating it into a trigger alert system capable of promptly warning medical staff of urgent situations through real-time capture and analysis of potentially harmful motions.