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

Background: The main classification systems of sleep disorders are based on the subjective self-reported criteria. Objective measures are essential to characterize the nocturnal sleep disturbance, identify daytime impairment, and determine the course of these symptoms. The aim of this study was to establish a resting-state fMRI-based support vector machine (SVM) classifier to diagnose insomnia disorder.

Methods: We enrolled 20 patients with insomnia disorder and 21 healthy controls, and obtained their simultaneous polysomnographic electroencephalography and functional magnetic resonance imaging (EEG-fMRI) recordings. The SVM classifiers were trained to capture insomnia. Classifier performance was quantified by a 5-fold cross validation and on independent test dataset.

Results: The fMRI-based SVM classifier was able to diagnose insomnia with an accuracy of 89.3% (sensitivity of 90.9%, specificity of 87.7%). The robustness of SVM classifier was encouraging.

Conclusions: We established an encouraging resting-state fMRI-based SVM classifier to automatically diagnose insomnia disorder. As an objective measure for assessing insomnia disorder, it would be of additional value to the current self-reported subjective criteria.