Method and system for subject-adaptive real-time sleep stage classification
US7509163B1 · kind B1 · utility
Assignee
Inventors
Key dates
| Filing date | Sep 28, 2007 |
| Grant date | Mar 24, 2009 |
| Priority date | — |
| Expiry date | Sep 28, 2027 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG16H50/70
- WIPO fieldMedical technology
- WIPO sectorInstruments
Abstract
A method of subject-adaptive, real-time sleep stage classification to classify electroencephalogram sleep recordings into sleep stages to determine whether a subject exhibits a sleep disorder includes performing subject adaptation to improve classification accuracy for a new subject with limited training data, the performing subject adaptation comprises using linear-chain conditional random fields and potential functions, training the linear-chain conditional random fields using the training data, continuously receiving the electroencephalogram waves, continuously extracting features from the electroencephalogram waves, the extracting features comprising transforming each of the electroencephalogram waves to capture information embedded in the electroencephalogram waves, and continuously classifying the sleep stages according to extracted features and learned parameters from the linear-chain conditional random fields.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.