Database Open Access
Preterm Infant Cardio-Respiratory Signals Database
Published: Feb. 9, 2017. Version: 1.0.0
A. H. Gee, R. Barbieri, D. Paydarfar and P. Indic, Predicting Bradycardia in Preterm Infants Using Point Process Analysis of Heart Rate, in IEEE Transactions on Biomedical Engineering, vol. 64, no. 9, pp. 2300-2308, Sept. 2017. doi:10.1109/TBME.2016.2632746.
Please include the standard citation for PhysioNet:
(show more options)
Goldberger, A., Amaral, L., Glass, L., Hausdorff, J., Ivanov, P. C., Mark, R., ... & Stanley, H. E. (2000). PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation [Online]. 101 (23), pp. e215–e220.
The Preterm Infant Cardio-respiratory Signals (PICS) database contains simultaneous ECG and respiration recordings of ten preterm infants collected from the Neonatal Intensive Care Unit (NICU) of the University of Massachusetts Memorial Healthcare. Statistical features based on linear estimates of heart rate are used to predict episodes of bradycardia.
Ten preterm infants were studied, with post-conceptional age of 29 3/7 to 34 2/7 weeks (mean: 31 1/7 weeks) and study weights of 843 to 2100 grams (mean: 1468 grams). The infants were spontaneously breathing room air and did not have any congenital or perinatal infection of the central nervous system, intraventricular hemorrhage of grade II or higher, and hypoxic-ischemic encephalopathy. A single channel of a 3-lead electrocardiogram (ECG) signal was recorded at 500 Hz (when available) from bedside patient monitors (Intellivue MP70, Philips Medical Systems) for ~20-70 hours per infant. In absence of an ECG channel, a compound ECG signal was recorded (250Hz). The compound ECG signal is an integrated signal of the three ECG lead channels. The choice of ECG availability was subject to nursing preference, and the researchers did not interfere with signals displayed on the Philips monitor.
Respiratory signals, using external inductance bands placed around the chest wall and abdomen, were also recorded (50 Hz) and synchronized using VueLoggerTM , a data acquisition system developed at the Wyss Institute, Harvard University. See the referenced paper for more details.
The single channel ECG and Respiration records are provided in standard WFDB format. Recording start times are synchronized within each infant. ECGs are recorded at 500Hz except for infants 1 and 5, which are recorded as a compound ECG signal at 250Hz. The ECG leads are specified in the header files. R-peaks are extracted from the ECGs using a modified Pan-Tompkins algorithm. The annotations are then visually inspected by researchers to remove artifacts due to movement, disconnection, and any erroneous peak detections.
Respiration signals are recorded at 50Hz from abdomen inductance bands, except for infant 1, which is recorded at 500Hz. The respiration signal for infant 1 was recorded using the respiration signal from the Philips monitor (i.e. abdomen inductance band information was not available at time of study). Respiration peaks are algorithmically extracted; however, these annotations have not been manually vetted yet.
Bradycardia onset annotations are also provided. We define bradycardia as events where the heart rate slows to less than 100 bpm (or equivalently R-R > 0.6 s) and for at least two beats (> 1.2 s) in duration. We also consider bradycardia clustering (i.e. successive bradycardias after a leading bradycardia). For our study, we aggregate any bradycardia within a 3-minute window of a leading bradycardia into one bradycardia to avoid statistical distortions during the prediction phase.
The ECG and respiration records are named in the form: infantN_ecg and infantN_resp, where N is the infant number.
Annotation files as described in the previous section are also provided, with record names matching the records they are associated with. The annotation file extensions are qrsc for ECG R locations, resp for respiration peaks, and atr for bradycardia onset times.
Anyone can access the files, as long as they conform to the terms of the specified license.
License (for files):
Open Data Commons Attribution License v1.0
Total uncompressed size: 1.6 GB.
Access the files
- Download the ZIP file (1.6 GB)
- Access the files using the Google Cloud Storage Browser here. Login with a Google account is required.
- Access the data using the Google Cloud command line tools (please refer to the gsutil documentation for guidance):
gsutil -m -u YOUR_PROJECT_ID cp -r gs://picsdb-1.0.0.physionet.org DESTINATION
- Download the files using your terminal:
wget -r -N -c -np https://physionet.org/files/picsdb/1.0.0/