HOPE-C: a dataset for high-risk observation and prevention evaluation using channel state information

This paper introduces the High-risk Observation and Prevention Evaluation - Channel State Information (HOPEC) dataset, a multi-modal resource designed for activity recognition and vital signs monitoring. The dataset features raw Channel State Information (CSI) signals annotated with activity classes that include simulated harmful behaviors, along with data from wearable devices capturing ECG, accelerometer, gyroscope, and respiratory signals. The dataset covers 27 sessions involving 23 participants, recorded in a simulated clinical environment resembling a secure care unit bedroom. Participants performed 12 activities typical of real-world scenarios during each session. Baseline experiments demonstrate the dataset’s validity, achieving 94.9% accuracy in distinguishing between movement and static classes. HOPE-C will be publicly released to facilitate the evaluation and comparison of health monitoring methods in high-risk clinical settings.