Ventiliser

Ventiliser segments and characterises neonatal ventilator airway pressure and flow waveforms sampled at 100 Hz from Dräger Babylog VN500 to identify inflations and sub-phases such as inspiration, expiration, pressure rise, pressure plateau, and inspiratory hold.


Key Features:

  • Input data: Processes airway pressure and flow waveforms sampled at 100 Hz from Dräger Babylog VN500 ventilators.
  • Implementation: Implemented as a Python package.
  • Automatic Segmentation: Employs a rule-based algorithm to discretize continuous ventilator data into distinct flow and pressure states and detect individual breaths.
  • Phase identification: Identifies sub-phases within each inflation cycle, including inspiration, expiration, pressure rise, pressure plateau, and inspiratory hold.
  • Accuracy: Detects over 97% of ventilator inflations and their sub-phases in out-of-sample validation against manually annotated datasets.
  • Processing performance: Processes up to 24 hours of ventilation data in approximately two minutes, with processing time increasing linearly for longer recordings.
  • Reporting: Generates detailed tables reporting indices for each breath and its sub-phases.
  • Visualization: Produces visualizations of individual inflations as waveforms or loops.
  • Longitudinal analysis: Enables quantitative analysis of ventilation patterns and ventilator–patient interactions over extended recordings.

Scientific Applications:

  • Clinical monitoring: Analysis of ventilation mechanics in neonatal intensive care to characterise ventilatory support in preterm and critically ill infants.
  • Research on ventilator–patient interaction: Quantitative study of ventilator–patient interactions and inflation dynamics over long-duration recordings.
  • Protocol and device evaluation: Comparative analysis of ventilation periods to support development and assessment of ventilation protocols and technologies.

Methodology:

Uses a rule-based algorithm together with an information gain approach to segment 100 Hz airway pressure and flow data from ventilators into inflations and sub-phases.

Topics

Details

License:
MIT
Programming Languages:
Python
Added:
1/18/2021
Last Updated:
3/11/2021

Operations

Publications

Chong D, Morley CJ, Belteki G. Computational analysis of neonatal ventilator waveforms and loops. Pediatric Research. 2020;89(6):1432-1441. doi:10.1038/s41390-020-01301-9. PMID:33288876. PMCID:PMC7720788.