37 2 Automated oxygen control in preterm infants, how does it work and what to expect Clinical effect CLiO2 is the most researched algorithm in clinical setting. 18, 28-33 All seven studies reported a significant increase in time spent within TR, including a reduction of28-30, 32, 33 or equal18, 31 time spent above it (Table 1). Three studies demonstrated a small but significant increase of mild hypoxia.28, 29, 33 But, in the largest multicentre trial31 the gain in TR time was mainly attributed to less hypoxia, rather than hyperoxia. There was heterogeneity in sample size, study duration and modes of respiratory support in the studies. The TR upper and lower limits also varied, this being relevant for the calculation of SpO2 error, thus influencing algorithm function. The CLiO2 algorithm has thus been seen to be superior to manual oxygen titration in a range of situations, including when used for longer periods of time as shown by van Zanten et al.33 In a recent meta-analysis38 on AOC, the effect on SpO 2 targeting was somewhat diminished when only including the CLiO2 studies compared to when all studies were included (8.9% vs 12.8% increase in time within TR). In our experience33, the CLiO 2 algorithm performs well in both stable and unstable preterm infants. However, there appears to be room for improvement, especially regarding the rapidity of FiO2 reduction in the event of clinical improvement, for example after surfactant therapy. The algorithm needs substantial time to adjust baseFiO2 downwards in this circumstance, which poses a problem as most FiO2 adjustments are limited to be ≥ baseFiO2. In addition, in unstable babies with frequent hypoxia the baseFiO2 does not reflect the basal oxygen requirement. AOC with the CLiO2 algorithm has been implemented in the NICU in Leiden University Medical Centre since mid-2015. A practical solution for the baseFiO2 problem has been to reset the algorithm when needed, which adopts the clinician-set FiO2 as baseFiO2. IntellO2 35 How it works The IntellO2 algorithm is a recent PID algorithm implemented in the Vapotherm Precision Flow device for delivering nasal high flow.35 Details on the algorithm are limited. It is reported to be a modified version of an earlier algorithm described by Bhutani.9 As with other algorithms, SpO 2 is measured by a built-in pulse-oximeter. If the SpO2 is lost for two minutes or is degraded by more than 50% in the last four minutes, the algorithm reverts to the back-up FiO2 value (the highest of a clinician-set back-up value or the median of the last three FiO2 values).