The application of machine learning approaches for the classification of people with migraine and neck pain versus neck pain only

Abstract

Migraine and neck pain are highly prevalent and disabling conditions and they typically present with overlapping symptoms. It remains unknown whether there are distinct clinical features that are able to discriminate between these two conditions. This is highly relevant since the management of neck pain differs to that of migraine thus identifying features which help clinicians to differentially diagnose these conditions is of high relevance. Objective: This project aims to investigate how supervised machine learning algorithms can discriminate individuals with migraine from those with idiopathic neck pain and from those that are asymptomatic. Features examined will include the quantity and quality of cervical range of motion, cervical endurance, submaximal isometric contractions, headache and neck-related disability, kinesiophobia, and catastrophizing. Methods: Individuals with migraine, idiopathic neck pain, and asymptomatic individuals will be assessed. The volunteers will complete questionnaires related to neck disability, headache disability, kinesiophobia, and pain catastrophizing followed by physical testing. Activity of the sternocleidomastoid, anterior scalene, splenius capitis, and upper trapezius will be recorded during all tests with electromyography. Expected results: We expect that the findings of this work may help differentiate patients with migraine and neck pain by relating them to the performance demonstrated by the evaluation of clinical characteristics, discriminating each group of patients. Establishing the clinical features that differentiate migraine and neck pain can subsequently demonstrate which individuals will benefit from different treatment techniques. (AU)