Modelling techniques and their application for monitoring in high dependency environments--learning models

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Artificial Intelligence in Medicine

Background:

• Biosignals, such as electroencephalograms (EEG) and evoked potentials (EP), are crucial for monitoring critical illness.
• Current analysis of these brain signals can detect early signs of inadequate brain perfusion.
• Learning models are increasingly used for biosignal analysis, often as 'black-box' systems.

Purpose of the Study:

• To review the application of learning models, including Bayesian classifiers and artificial neural networks, for biosignal interpretation.
• To highlight the potential of using brain signals for early detection of central nervous system damage.
• To advocate for the integration of biosignal monitoring into critical care databases.

Main Methods:

• Review of learning models (Bayesian classifiers, artificial neural networks) for biosignal analysis.
• Focus on interpretation of electrical brain signals (EEG, EP).
• Discussion of signal changes indicating inadequate brain perfusion.

Main Results:

• Learning models follow similar training and application sequences.
• Sudden changes in EEG or EP are early indicators of brain perfusion issues.
• These issues can stem from systemic failures like oxygen desaturation or hypotension.

Conclusions:

• Brain signals (EEG, EP) recorded in critical care units are valuable for early warning systems.
• Integrating these signals into annotated databases (e.g., IMPROVE project) can facilitate new detection methods.
• On-line monitoring can enable timely interventions to prevent permanent central nervous system damage.