During general anesthesia, intraoperative hypotension (IOH) is associated with increased morbidity and mortality. Mean arterial pressure (MAP) < 65mmHg is the most common definition of hypotension. In order to reduce IOH, a complex method using machine learning called hypotensive prediction index (HPI) was shown to be superior to changes in MAP (ΔMAP) to predict hypotension (MAP between 65 and 75 excluded). Linear extrapolation of MAP (LepMAP) is also very simple and could be a better approach than ΔMAP. The main objective of the present study was to investigate whether LepMAP could predict IOH during anesthesia 1, 2 or 5 minutes before.

Hypothesis : the area under the ROC curves (ROC AUCs) at 1, 2 and 5 minutes of LepMAP would be superior to AMAP

Inclusion Criteria:

• Patients suffering from sleep pathologies (apneas, narcolepsy, etc.) hospitalized for polysomnography and MWT, in order to evaluate their sleepiness (usually before they resume driving)
• hospitalized in the Center for Sleep Medicine (Hôpital Lyon Croix Rousse) between September 2017 and January 2019

Exclusion Criteria:

• patient with active encephalopathy or epilepsy whose EEG was abnormal
• denial of study participation
• availability of all the recordings (n=4 tests) of the MWT

Patients suffering from sleep pathologies (apneas, narcolepsy, etc.) hospitalized for polysomnography and wakefulness tests, generally with a view to authorizing driving.