Functional connectivity-based classification of rapid eye movement sleep behavior disorder.

Area of Science:

• Neuroscience
• Medical Imaging
• Machine Learning

Background:

• Isolated rapid eye movement sleep behavior disorder (iRBD) is a key indicator of future synucleinopathies.
• Developing methods to assess iRBD's underlying brain states is crucial for early detection.
• Resting-state fMRI and machine learning offer a promising avenue for this evaluation.

Purpose of the Study:

• To develop and validate a machine learning classifier for identifying iRBD using functional connectivity.
• To explore the relationship between functional connectivity patterns and clinical symptoms in iRBD patients.

Main Methods:

• A machine learning classifier was built using functional connectivity data from 55 iRBD patients and 97 healthy controls (HC).
• Random forest identified important functional connectivities, which were then used for classification via logistic regression and SVM.
• Classification performance was evaluated, and correlations between connectivity and clinical symptoms were analyzed.

Main Results:

• The classifier achieved significant accuracy (0.649), precision (0.659), recall (0.662), and F1 score (0.645), with an AUC of 0.707.
• Key functional connectivities involved motor, somatosensory, parietal, temporal, thalamic, and cerebellar areas.
• Variations in functional connectivity correlated with subclinical motor and non-motor symptoms in iRBD patients.

Conclusions:

• Machine learning classifiers utilizing functional connectivity show potential for evaluating latent brain states in iRBD.
• This technology could aid in the early detection and management of synucleinopathies.