Laboratory:
Swiss Data Science Center

Type:
Semester Project

Description:
Polysomnography (PSG) is the current gold standard for sleep monitoring and clinical assessment of sleep disorders. However, this technique is intrusive, in general limited to a single night in-hospital of recordings, and therefore not representative of “natural sleep” at home. Alternative, inexpensive, and non-intrusive monitoring is thus needed for accurate and recurrent sleep monitoring. In the project, we aim to predict, using machine learning algorithms, sleep-stages and potential sleep disorders, from pressure sensors recordings located in a mat.

Goals/benefits:

• Getting a first experience in supervised machine learning.
• Experience working with a real data set from medical sciences.
• Possibility to participate to the development of a start-up.

Prerequisites:

• Courses in machine learning and statistics (time series analysis).
• Basic knowledge of Python/R and corresponding packages (scikitlearn, pandas, . . . ).
• Knowledge of Jupyter Notebooks/RMarkdown.

Deliverables:

• Clean and documented code
• Report
• Oral presentation

Contact:
Raphaël de Fondeville: [email protected]