Discriminating Paradoxical and Psychophysiological Insomnia Based on Structural and Functional Brain Images: A Preliminary Machine Learning Study

Abstract

Insomnia disorder (ID) is a prevalent mental illness. Several behavioral and neuroimaging studies suggested that ID is a heterogenous condition with various subtypes. However, neurobiological alterations in different subtypes of ID are poorly understood. We aimed to assess whether unimodal and multimodal whole-brain neuroimaging measurements can discriminate two commonly described ID subtypes (i.e., paradoxical and psychophysiological insomnia) from each other and healthy subjects. We obtained T1-weighted images and resting-state fMRI from 34 patients with ID and 48 healthy controls. The outcome measures were grey matter volume, cortical thickness, amplitude of low-frequency fluctuation, degree centrality, and regional homogeneity. Subsequently, we applied support vector machines to classify subjects via unimodal and multimodal measures. The results of the multimodal classification were superior to those of unimodal approaches, i.e., we achieved 81% accuracy in separating psychophysiological vs. control, 87% for paradoxical vs. control, and 89% for paradoxical vs. psychophysiological insomnia. This preliminary study provides evidence that structural and functional brain data can help to distinguish two common subtypes of ID from each other and healthy subjects. These initial findings may stimulate further research to identify the underlying mechanism of each subtype and develop personalized treatments for ID in the future.

Keywords: classification; insomnia disorder; machine learning; multimodal imaging; paradoxical insomnia; psychophysiological insomnia.

Figure 1

Multimodal confusion matrix (green color: correctly classified; red color: incorrectly classified) of HC vs. PPI classification shows 40 HCs and 14 PPIs were correctly classified (A). Confusion matrix of HC vs. PDI classification shows 44 HCs and 11 PDs were correctly classified (B). The confusion matrix of PDI vs. PPI classification shows 12 PDIs and 17 PPIs were correctly classified (C). The most important features (parcels) in each classification were illustrated: (D) HC vs. PPI, (E) HC vs. PDI, and (F) PDI vs. PPI. Important features were selected by analyzing each feature’s amplitude of the eigenvectors. The color bar indicates percent values of the amplitude of eigenvector for each feature. Cortical parcels were extracted by the Schaefer brain atlas and subcortical ones were extracted based on the Brainnetome atlas. HC: healthy control, PDI: paradoxical insomnia, PPI: psychophysiological insomnia.