Prospective Observational Study of Automated Apnoea Detection in Preterm Infants Receiving Non-invasive Respiratory Support
Prospective Observational Study of Automated Apnoea Detection in Preterm Infants Receiving Non-invasive Respiratory Support
The aim of this study is to monitor the frequency of apnoeas (pauses in breathing) on various methods of non-invasive respiratory support that are detected by an automated machine-learning (ML) model based on diaphragmatic electromyography (dEMG), in infants born at less than 32 weeks of gestation.
Our hypothesis is that the ML algorithm will improve identification of apnoeic episodes and their classification to central or obstructive.
The study will measure outcomes including the number of apnoeic episodes during the monitoring period, their classification to central and obstructive apnoeas and the predictive ability of the machine-learning algorithm to correctly identify and classify these episodes compared to those documented in nursing charts. Correct classification of apnoeic episodes may help identify underlying causes that require specific intervention.
When the clinical team identifies an infant is eligible for enrolment to the study and following verbal assent of the attending neonatal consultant, a member of clinical staff will initially approach the parents/ legal guardians of eligible infants and if they agree, a researcher. The parents will be provided with an information sheet about the study. The researchers will answer questions and respond to any concerns in a face-to-face meeting. Written informed consent will be obtained.
Electrical activity of the diaphragm, airway pressure, flow and peripheral oxygen saturation levels will be recorded for a duration of eight hours. Transcutaneous diaphragm EMG (sEMG) will be monitored using three surface electrodes (3M Red Dot Foam monitoring electrode 2228, 3M, United Kingdom) that are placed on the infant's abdomen and sternum. The electrodes are connected to a small battery-operated measuring device (SERA, DEMCON; Makawi Medical Systems, the Netherlands) that amplifies and pre-processes the signals received from the electrodes. The pre-processed signals are sent via a Bluetooth connection to a receiving unit that performs higher level processing to derive the EMG signal and other measurements. These results are communicated via a wired connection to a bedside computer running SERA Graphical User Interface (GUI) software.
Airway pressure and flow signals will be measured by a flow sensor and pressure tube (Sensirion AG, Stäfa, Switzerland) that will be placed between the exit of the SLE6000/ SLE6000N ventilators (Inspiration Healthcare, Croydon, UK) and the tubing that is attached to the patient. This will ensure pressure and flow data are recorded simultaneously with the sEMG signal. These data will also be fed to the SERA measuring device.
An SpO2 cable (SLE uSpO2, Inspiration Healthcare, Croydon, UK) will be connected to the patient and the ventilator for continuous (second by second) recording of SpO2 levels. These data will be time synced using post processing.
Participants will also be connected to the standard bedside monitor (Phillips Intellivue MX750) for the whole duration of the study.
Surface EMG, pressure, flow and SpO2 data will be synchronised. Two researchers will identify all apnoeic episodes and classify them as central, obstructive or noise.
The ML algorithm will also be used to identify and classify apnoeic episodes that occurred during the study duration.
Comparisons will be made between the apnoeic episodes identified the researchers, the ML algorithm and the nursing staff electronic patient records. These records include classification of an episode as apnoea or desaturation or bradycardia, its duration and any actions taken.
Inclusion Criteria:
Preterm infants <32 weeks of gestation at birth and up to 36 weeks postmenstrual age, on non-invasive respiratory support including:
Exclusion Criteria:
ourania.kaltsogianni@nhs.net0044+02032999000 ext. 38494