Extract

It is becoming increasingly apparent that the apnoea/hypopnoea index does not reflect the complex physiological processes associated with obstructive sleep apnoea. Consequently, the use of this metric in isolation is unlikely to accurately predict cardiovascular risk [1, 2]. As an alternative, more comprehensive and combined metrics are necessary to clearly define cardiovascular risk in patients living with obstructive sleep apnoea. The work of Martinot et al. [3], highlighted in the present issue of the European Respiratory Journal, addresses this concern by examining the complex interaction of numerous parameters in predicting daytime hypertension. To examine this relationship, a machine learning algorithm was optimised and validated before being used to determine the respective power of association of 18 different parameters with hypertension.