The novel PID algorithm was very effective for automated oxygen control in preterm infants, and deserves further investigation.

An enhanced PID algorithm was very effective for automated oxygen control in a simulation of oxygenation, and deserves clinical evaluation.

Automated control of inspired oxygen for newborn infants is an emerging technology, currently limited by reliance on a single input signal (oxygen saturation, SpO). This is while other signals that may herald the onset of hypoxic events or identify spurious hypoxia are not usually utilised. We wished to assess the frequency of apnoea, loss of circuit pressure and/or motion artefact in proximity to hypoxic events in preterm infants on non-invasive ventilation. Hypoxic events (SpO < 80 %) were identified using a previously acquired dataset obtained from preterm infants receiving non-invasive ventilation. Events with concomitant apnoea, loss of circuit pressure or oximetry motion artefact were annotated, and the frequency of each of these factors was determined. The effect of duration and timing of apnoea on the characteristics of the associated hypoxic events was studied. Among 1224 hypoxic events, 555 (45 %) were accompanied by apnoea, 31 (2.5 %) by loss of circuit pressure and 696 (57 %) by motion artefact, while for 224 (18 %) there were no concomitant factors identified. Respiratory pauses of longer duration (>15 s) preceding hypoxic events, were associated with a relatively slow decline in SpO and more prolonged hypoxia compared to shorter pauses. Hypoxic events are frequently accompanied by respiratory pauses and/or motion artefact. Real-time monitoring and input of respiratory waveform may thus improve the function of automated oxygen controllers, allowing pre-emptive responses to respiratory pauses. Furthermore, use of motion-resistant oximeters and plethysmographic waveform assessment procedures will help to optimise feedback control of inspired oxygen delivery.