Page 220 - MEGIN Book Of Abstracts - 2023

Migraine

Resting-state magnetoencephalographic oscillatory 89.1%, AUC: 0.91) in classifying CM from other pain
connectivity to identify patients with chronic disorders (FM in this study). These resting-state magne-
migraine using machine learning (2022) toencephalographic electrophysiological features yield
oscillatory connectivity to identify patients with CM
Hsiao, Fu-Jung; Chen, Wei-Ta; Pan, Li-Ling Hope; Liu, from those with a different type of migraine and pain
Hung-Yu; Wang, Yen-Feng; Chen, Shih-Pin; Lai, Kuan-Lin; disorder, with adequate reliability and generalizability.
Coppola, Gianluca; Wang, Shuu-Jiun
Keywords: Chronic migraine, Machine learning, Magneto-
Brain Research Center, National Yang Ming Chiao Tung Uni- encephalography, Pain disorders, Resting-state oscillatory
versity, Taipei, Taiwan. [email protected]; Department connectivity
of Neurology, Keelung Hospital, Ministry of Health and Wel-
fare, Keelung, Taiwan. [email protected]; Department The journal of headache and pain (2022), Vol. 23, No. 1
of Neurology, Neurological Institute, Taipei Veterans General (36192689) (0 citations)
Hospital, 201, Shih Pai Rd Sec 2, Taipei, Taiwan, 11217; De-
partment of Medico-Surgical Sciences and Biotechnologies,
Sapienza University of Rome Polo Pontino, Latina, Italy Headache-related circuits and high frequencies
evaluated by EEG, MRI, PET as potential biomarkers
ABSTRACT To identify and validate the neural signa- to differentiate chronic and episodic migraine:
tures of resting-state oscillatory connectivity for chronic Evidence from a systematic review (2022)
migraine (CM), we used machine learning techniques
to classify patients with CM from healthy controls (HC) Gomez-Pilar, Javier; Martínez-Cagigal, Víctor; García-
and patients with other pain disorders. The cross- Azorín, David; Gómez, Carlos; Guerrero, Ángel; Hornero,
sectional study obtained resting-state magnetoen- Roberto
cephalographic data from 240 participants (70 HC, 100
CM, 35 episodic migraine [EM], and 35 fibromyalgia Centro de Investigación Biomédica en Red en Bioingeniería,
[FM]). Source-based oscillatory connectivity of relevant Biomateriales Y Nanomedicina (CIBER-BBN), Valladolid,
cortical regions was calculated to determine intrinsic Spain; Headache Unit, Neurology Department, Hospital
connectivity at 1-40 Hz. A classification model that Clínico Universitario de Valladolid, Ramón y Cajal 3, 47003,
employed a support vector machine was developed Valladolid, Spain. [email protected]; Department of Medi-
using the magnetoencephalographic data to assess the cine, University of Valladolid, Valladolid, Spain
reliability and generalizability of CM identification. In
the findings, the discriminative features that differenti- BACKGROUND The diagnosis of migraine is mainly
ate CM from HC were principally observed from the clinical and self-reported, which makes additional ex-
functional interactions between salience, sensorimotor, aminations unnecessary in most cases. Migraine can be
and part of the default mode networks. The classifica- subtyped into chronic (CM) and episodic (EM). Despite
tion model with these features exhibited excellent the very high prevalence of migraine, there are no
performance in distinguishing patients with CM from evidence-based guidelines for differentiating between
HC (accuracy ≥ 86.8%, area under the curve (AUC) ≥ 0.9) these subtypes other than the number of days of mi-
and from those with EM (accuracy: 94.5%, AUC: 0.96). graine headache per month. Thus, we consider it timely
The model also achieved high performance (accuracy: to perform a systematic review to search for physiologi-

 ontents  Index 199
C