Insomnia disorder diagnosed by resting-state fMRI-based SVM classifier.

Area of Science:

• Neuroimaging
• Sleep Medicine
• Machine Learning

Background:

• Current insomnia diagnosis relies on subjective self-reported criteria.
• Objective measures are crucial for characterizing sleep disturbances and daytime impairment.
• There is a need for objective diagnostic tools in sleep disorders.

Purpose of the Study:

• To develop a resting-state functional magnetic resonance imaging (fMRI) based support vector machine (SVM) classifier for diagnosing insomnia disorder.
• To establish an objective method for insomnia diagnosis.

Main Methods:

• Recruited 20 insomnia patients and 21 healthy controls.
• Acquired simultaneous electroencephalography (EEG) and fMRI recordings.
• Trained SVM classifiers using resting-state fMRI data to identify insomnia.

Main Results:

• The fMRI-based SVM classifier achieved 89.3% accuracy in diagnosing insomnia.
• Demonstrated high sensitivity (90.9%) and specificity (87.7%).
• The classifier's robustness was confirmed through 5-fold cross-validation and an independent test dataset.

Conclusions:

• An effective resting-state fMRI-based SVM classifier for automatic insomnia diagnosis was established.
• This objective measure complements existing subjective diagnostic criteria.
• The developed classifier shows promise as an objective tool for assessing insomnia disorder.