Machine Learning-Derived Neural Signatures of Itch and Pain That Reliably Distinguish the Two Sensations in Humans: A...

Area of Science:

• Neuroimaging
• Machine Learning
• Biomarker Development

Background:

• Neuroimaging and machine learning identify neural signatures of internal states.
• No prior studies have identified neural signatures distinguishing itch and pain.
• This study explores itch and pain neural signatures using functional MRI (fMRI) and support vector machine (SVM).

Purpose of the Study:

• To identify and distinguish neural signatures of itch and pain.
• To explore the potential for developing objective biomarkers for itch and pain.

Main Methods:

• fMRI data collected from 33 participants under itch, pain, and control conditions.
• Seed-based functional connectivity images (R-images) created using posterior cingulate cortex (PCC) and anterior insular cortex (aIC).
• Cross-validated and bootstrapped SVM applied to R-images to identify key brain regions differentiating itch and pain.

Main Results:

• Neural signatures demonstrated excellent classification performance (>90% accuracy, >0.9 AUC).
• Combined signatures showed particularly high classification capability.
• Identified key brain regions contributing to the distinction between itch and pain.

Conclusions:

• First neuroimaging study to identify reliable neural signatures for itch and pain using machine learning.
• Demonstrated high classification performance through seed-based functional connectivity and SVM.
• Presents a proof-of-concept for developing brain-based biomarkers for itch and pain assessment.