Thesis Project Form

Understanding sleep stages is critical for diagnosing and managing sleep disorders, which affect health and well-being, particularly for certain categories of subjects affected by sleep disorders.
Electroencephalograms (EEG) and hypnograms are key tools for assessing sleep, but manual analysis is time-consuming and prone to variability. This thesis bridges bioengineering and AI, exploiting machine learning to automate sleep stage classification, aiming for improved accuracy and efficiency.

The general objectives of this thesis are:

Automate Sleep Behavior Analysis: Develop a machine learning pipeline to automate the classification of sleep stages from polysomnographic (PSG) recordings, reducing the time and effort required for manual analysis. In particular:

Feature Extraction for Shallow Learning: Identify and extract meaningful features to train shallow learning models, ensuring accurate and interpretable classification.
Deep Learning: Leverage deep learning techniques to directly process raw signals, aiming to optimize feature extraction and classification in an end-to-end framework.

Experimental Design for Robustness: Implement advanced resampling strategies, such as cross-validation and bootstrapping, to evaluate model performance and ensure robustness and generalizability across diverse datasets.

-data handling and preparation
- medical imaging data preprocessing
- feature extraction
- deep and shallow learning models