Overview This dataset contains ground truth physiological values and processed RGB-D video-derived signals for the non-contact estimation of vital signs in neonates. Data were collected in neonatal intensive care units using RGB-D cameras and clinical monitors. The dataset is intended for use in evaluating and developing AI and computer vision algorithms for contactless vital sign monitoring.

Data Structure The dataset is organized by patient, with each patient assigned a unique folder under the main dataset directory (e.g., mainfolder/mk501/). Within each patient folder, there are one or more subfolders, each corresponding to a different data modality (e.g., ECG, HeartRate, Respiration, etc.).

Each modality folder contains a single .csv file with time-series data. All data streams are time-aligned and use a standardized timestamp format, with time zero corresponding to the start of the individual recording session for that patient.

File Format and Metadata

Each .csv file contains a time series of a single physiological or video-derived variable.

Columns indicate the recorded variable values, with units provided in the header where appropriate.

Timestamps are standardized and represent seconds elapsed since the beginning of the recording session.

Empty modality folders indicate that no data were recorded for that modality in the corresponding session.

Citation If you use this dataset in your research or publications, please cite:

Ruhrberg Estévez, S., Grafton, A., Thomson, L., Warnecke, J., Beardsall, K., & Lasenby, J. (2025). Continuous non-contact vital sign monitoring of neonates in intensive care units using RGB-D cameras. Scientific Reports, 15(1). https://doi.org/10.1038/s41598-025-00539-9