Machine Learning-Based Analysis of Salivary Proteomic Signatures in Catathrenia

Area of Science:

• Sleep Medicine
• Proteomics
• Machine Learning

Background:

• Catathrenia is a rare sleep disorder involving groaning, often misdiagnosed as sleep apnea.
• Understanding catathrenia's etiology is crucial for accurate diagnosis and treatment.

Purpose of the Study:

• To identify salivary protein biomarkers for catathrenia using proteomics and machine learning.
• To explore the potential link between catathrenia and extracellular matrix (ECM) pathways.

Main Methods:

• 4D-DIA proteomics analyzed salivary samples from 22 catathrenia patients and 22 controls.
• Machine learning selected potential protein markers, validated in a separate cohort.
• Polysomnography and Epworth Sleepiness Scale (ESS) were used for evaluation.

Main Results:

• Significant differences in salivary protein profiles were found between catathrenia patients and controls.
• Five potential protein markers (HATL5, B2RBF5, A0A7S5BZF8, PRR27, BCAT2) were identified.
• Protein marker abundance correlated with ESS scores and polysomnographic parameters.

Conclusions:

• Salivary proteomics and machine learning can identify potential biomarkers for catathrenia.
• Catathrenia may involve ECM-related airway rigidity during sleep.
• Further research is needed to confirm these findings and elucidate mechanisms.