Modélisation sensori-motrice des états de transition

The neural mechanisms underlying states of alertness have been the subject of research by several generations of researchers and physicians, but they remain largely unexplained. Furthermore, anaesthetists practice daily to modulate these same states of alertness to enable surgical procedures to be performed. During anaesthesia, in addition to the depth of sedation, the homeostasis of the major physiological functions must be monitored because it can be disturbed. In current practice, the monitoring of cardiorespiratory parameters and the doses of injected anaesthetics are therefore used to have an indirect reflection of the sedation. In this context, the work of our thesis has a twofold aim: firstly, to deepen the knowledge of the mechanisms underlying the modulation of states of alertness; secondly, to search for methods that can monitor in real time the level of consciousness of patients during surgery. To this end, we have attempted to complement the anaesthetist's assessment with an algorithm that would allow an objective estimation of the state of consciousness from the monitoring of non-EEG physiological data available during anaesthesia in current practice. The conclusions of this algorithm were then compared to those provided by the anaesthetist and to data collected simultaneously by a 32-channel EEG. Using our databases, an algorithm was developed to predict the depth of anaesthesia without incorporating EEG data. This method, based on a hidden Markov model, has been the subject of a patent application, an accepted publication and another in the process of being submitted. The algorithm can determine the state of a patient during surgery with a true positive rate of 75%. In order to improve this prediction rate, several avenues are possible, including the integration of EEG signals into the data used by the algorithm. However, in a clinical practice context, we have already mentioned that the use of a multi-channel EEG is not realistic. However, by analysing data from 32 EEG channels during general anaesthesia, we concluded that recording a single channel at the frontal level was sufficient to discriminate a patient's level of anaesthesia depth, and this has been published. Our monitoring of the modulation of the states of consciousness of anaesthetised patients continued until the third hour after the end of anaesthesia. Our aim was to investigate when the EEG returned to the awake state. Our preliminary results show that during the early awakening phase, there were homogeneous awakening trajectories across the patients' scalps. However, the following 3 hours were marked by major inter-individual differences in the disappearance of brain waves characteristic of anaesthesia. The significance of these phenomena is still unclear and is the subject of further work. In conclusion, collecting several databases during anaesthesia has allowed us to open up new avenues of research and our future work will therefore continue along three lines i) the improvement of the algorithm that determines the depth of anaesthesia, ii) the implementation of a follow-up of patients for several hours after anaesthesia, until recovery ad integrum and obtaining an EEG trace strictly identical to what it was before the anaesthesia, and iii) the application of the new methodologies of monitoring the depth of sedation in intensive care unit.