Evaluating Non-Invasive Spontaneous Breath Detection in Preterm Infants: Development and Assessment of a Data Extraction Method to Clinically Validate the Pressure Trigger Algorithm in the SLE 6000 Ventilator

Evaluating Non-Invasive Spontaneous Breath Detection in Preterm Infants: Development and Assessment of a Data Extraction Method to Clinically Validate the Pressure Trigger Algorithm in the SLE 6000 Ventilator

Blom, Fabio (TU Delft Mechanical, Maritime and Materials Engineering)

te Pas, Arjan (mentor)
Cramer, Sophie (graduation committee)
Brouwer, Fleur (graduation committee)

Delft University of Technology

Technical Medicine

2023-10-13

The pressure trigger (PT) algorithm of the SLE 6000 ventilator is a non-invasive respiratory rate (RR) monitoring technique that can detect spontaneous breathing during non-invasive respiratory support. The PT algorithm can detect spontaneous breaths based on air pressure changes in the ventilatory circuit, labelling the pressure fluctuation as a PT. The threshold for the algorithm to detect a PT is set by the trigger sensitivity (TS), a user-set variable ranging between 0-100%. Although regularly used in the neonatal intensive care (NICU) of the Leiden University Medical Centre (LUMC), the PT algorithm still needs to be clinically validated as its performance in spontaneous breath detection is still unknown. To investigate the accuracy of the PT algorithm, it is necessary to collect timestamps of detected PTs, but these are not yet accessible. The aim of this study was to design and validate an extraction method to obtain the PT timestamps. For this, an experimental setup (EXP-SET) was designed, which connects to the ventilatory circuit of the SLE 6000 and allows us to perform to acquire the necessary timestamps. A replicated version of the PT algorithm detects PT timestamps from the raw pressure data acquired from the EXP-SET. To validate the similarity in PT detection between the EXP-SET and the SLE 6000, standardized and unstandardized bench tests were performed. The SLE 6000 was assumed to be the golden standard in PT detection during the bench tests. In the standardized bench test, the EXP-SET showed a sensitivity between 0.952 - 1.0 and a positive predictive value (PPV) between 0.997 - 1.0 for 50% TS. However, when increasing TS to 100%, the EXP-SET’s sensitivity and PPV decreased to 0.466- 0.618 and 0.673 - 0.844. In the unstandardized bench test, the sensitivities and PPVs were 0.794 and 1 for 50% TS, 0.901 and 0.996 for 75% TS and 0.785 and 0.958 for 100% TS. The results show that the designed EXP-SET does not yet detect PTs identically to the SLE 6000. The most probable cause is the different sampling frequency used in the EXP-SET compared to the SLE 6000, which leads to different behavior between the PT algorithm and replicated algorithm. Nevertheless, if the EXP-SET is further improved, it shows potential as an alternative acquisition method for the timestamp collection of spontaneous breaths for the SLE 6000. A properly functioning EXP-SET will enable clinical validation of the PT algorithm in the SLE 6000.

Preterm Infants
Non-invasive respiratory support
Spontaneous breathing
Pressure trigger

http://resolver.tudelft.nl/uuid:cb35f120-a8e1-4166-aafa-2f908f235685

Student theses

master thesis

© 2023 Fabio Blom