Kinematic Training in Patients With Neck Pain Based on Machine Learning Classification Approach

Overview

The goal of this clinical trial is to study if kinematic training based on novel kinematic assessment clasification approach can decrease chronic neck pain and prevent its reoccurance better than conventional kinematic training in adults. The main question\[s\] it aims to answer \[is/are\]: Does clustering patients with neck pain based on head and neck movement characteristics lead to more efficient kinematic rehabilitation training and improved clinical outcomes Researchers will compare effects of cluster specific kinematic training to see if it effects pain levels and its reoccurance. Participants will \[describe the main tasks participants will be asked to do, interventions they'll be given and use bullets if it is more than 2 items\].

Chronic idiopathic neck pain represents one of the most prevalent musculoskeletal conditions worldwide and is characterized by recurrent episodes, fluctuating symptom intensity, and frequent transition to persistent disability. While clinical management commonly focuses on short-term pain reduction, the more substantial societal burden arises from recurrence and chronicity rather than isolated acute episodes. Recurrent neck pain contributes to repeated healthcare consultations, prolonged work absenteeism, reduced productivity, and increasing healthcare expenditure. Despite the widespread use of exercise-based rehabilitation, recurrence rates remain high, suggesting that prevailing treatment approaches may insufficiently address the mechanisms underlying persistent dysfunction. As the prevalence of neck pain continues to rise across working-age populations, it is reasonable to question whether current rehabilitation models adequately target the neuromuscular and sensorimotor contributors to chronicity and repeated symptom exacerbations. Most individuals presenting for physical therapy are diagnosed with non-specific or idiopathic neck pain, a broad category encompassing heterogeneous clinical presentations with variable combinations of pain, stiffness, dizziness, visual disturbances, and impaired movement control. Patients may differ substantially in movement strategies, proprioceptive acuity, neuromuscular coordination, and adaptive motor patterns, yet they are frequently managed using standardized exercise protocols. Conventional clinical assessment tools are effective in distinguishing patients from asymptomatic individuals but provide limited resolution for identifying meaningful subgroups within the idiopathic neck pain population. Over the last decade, patient-centered evaluation in spinal disorders has frequently relied on numerical pain rating scales and disability questionnaires, while comparatively less emphasis has been placed on objective assessment of neuromuscular control mechanisms that may drive chronicity. Emerging evidence demonstrates that individuals with a history of neck pain exhibit persistent alterations in sensorimotor function even during symptom remission, implying that underlying neuromuscular adaptations may predispose individuals to recurrence and reduced resilience to mechanical or psychosocial stressors. Cervical sensorimotor control, often conceptualized as cervicocephalic kinaesthetic awareness, integrates proprioceptive input from cervical musculature and joint mechanoreceptors with visual and vestibular information to coordinate accurate head positioning and smooth eye-head movements. Altered cervical afferent input can disrupt this integration, leading to inaccurate movement perception, impaired position sense, and maladaptive motor output strategies. Prolonged proprioceptive disturbances may induce central neural plasticity changes within brainstem and cortical sensorimotor networks, potentially sustaining dysfunction beyond the initial pain episode and contributing to the maintenance of chronic symptoms. Systematic reviews report variable findings regarding the magnitude of kinaesthetic impairments in neck pain populations, likely reflecting recruitment heterogeneity, differences in symptom duration, and variability in functional deficits. These inconsistencies underscore the importance of multidimensional characterization of sensorimotor performance rather than reliance on isolated parameters or single test outcomes. Kinaesthetic training targeting movement accuracy, proprioceptive recalibration, and coordinated motor output has demonstrated clinically meaningful short-term improvements in pain and disability, with some effects persisting for several months. However, approximately one third of patients fail to demonstrate improvement in pain intensity or disability, and nearly half do not exhibit measurable gains in objective kinematic performance variables. Additionally, dropout rates of up to twenty percent have been reported, frequently attributed to discomfort, fatigue, or symptom aggravation during exercises that challenge impaired sensorimotor systems. These observations suggest that uniform rehabilitation protocols may not optimally address individualized neuromuscular profiles and that subgroup-specific interventions may enhance treatment responsiveness, improve adherence, and reduce variability in outcomes. Cervical movement control involves coordinated interaction between proprioceptive, vestibular, and visual systems. Movement tasks performed at varying amplitudes and velocities challenge different components of sensorimotor processing and may reveal distinct deficit patterns. Slow, large-amplitude movements increase reliance on proprioceptive discrimination and smoothness regulation, often reflected in elevated jerk index values when coordination is impaired. Faster movements may engage vestibular-dependent mechanisms and dynamic stability control. Furthermore, altered afferent input from cervical muscles can influence adaptive plastic changes in vestibular-dependent motion sensitivity, potentially affecting head and neck control at higher movement velocities. Consequently, assessment and training should incorporate movements across multiple amplitudes and velocities to adequately capture subgroup-specific impairments. Head and neck movement control tests frequently require participants to track visually presented targets while performing controlled movements, thereby demanding coordinated eye and head interaction. The importance of the visual system in neck pain is supported by neurophysiological connections between upper cervical afferent pathways and visual and vestibular nuclei at the brainstem level. Patients with neck pain frequently present with oculomotor disturbances, particularly during neck torsion maneuvers, likely arising from altered cervico-ocular and cervico-collic reflex contributions. Previous work has suggested associations between cervical kinaesthetic deficits and eye movement control, yet it remains unclear to what extent targeted kinaesthetic training can modify these oculomotor adaptations and whether such changes parallel improvements in pain and functional status. Traditional analytical strategies commonly evaluate individual kinematic parameters independently, potentially overlooking complex interactions among movement precision, amplitude regulation, and smoothness. High-dimensional datasets generated from movement control testing contain non-linear relationships that may not be adequately captured using conventional univariate statistical methods. Data mining and machine learning approaches enable identification of latent structure within multidimensional datasets and facilitate clustering of patients according to shared functional characteristics. Prior analyses conducted on a cohort of 135 individuals with idiopathic neck pain identified four distinct clusters based on combinations of movement control variables, pain intensity, and demographic characteristics. These clusters demonstrated unique kinematic profiles, supporting the existence of clinically meaningful heterogeneity within this population and suggesting that targeted therapeutic strategies may be warranted. The present study applies a classification-based rehabilitation framework derived from these previous clustering findings. Cervical kinaesthetic function will be comprehensively assessed using movement control testing and head-to-neutral relocation testing, supplemented by demographic and clinical variables including pain intensity. Participants will be assigned to previously defined clusters using established classification algorithms that integrate multidimensional movement and demographic data. Following classification, participants will receive either cluster-specific kinematic training designed to target the predominant impairments characteristic of their subgroup or a general kinematic training program not tailored to cluster membership. The structure, frequency, and duration of training exposure will be comparable between groups to ensure that differences in outcome are attributable to content rather than dose. Three-month follow-up measurements will be conducted to evaluate the persistence of intervention effects and to explore whether individualized rehabilitation influences medium-term symptom stability. Movement control assessment will involve visually guided head tracking tasks performed with inertial measurement units capturing high-resolution angular displacement data in real time. Participants will complete repeated trials across four progressively challenging movement paths characterized by increasing target amplitude and velocity to systematically stress different components of the sensorimotor system. Derived parameters will include precision time, representing percentage of trial time within the target zone; underreaching and overreaching, representing directional amplitude control errors; jerk index, reflecting smoothness of movement derived from the third derivative of position data; and amplitude accuracy, calculated as the average angular mismatch between target and performed movement. Position sense will be evaluated through blindfolded relocation to a self-selected neutral head position, with absolute error computed across repetitions and movement directions to quantify repositioning accuracy. Oculomotor control will be assessed using smooth pursuit testing in neutral and torsion positions, with eye movement recordings filtered for artifacts, synchronized with reference trajectories, and analyzed to compute pursuit gain as the ratio of eye velocity amplitude to target velocity amplitude. Cervical range of motion will be quantified in all primary movement planes to characterize mobility profiles and potential mobility-related subgroup distinctions. Data processing will involve systematic filtering of motion and eye movement signals, removal of blinks and saccades where appropriate, synchronization of reference trajectories, normalization of movement cycles, and averaging across repetitions to ensure stable parameter estimation. Statistical analyses will be performed using repeated-measures analysis of variance to evaluate time-by-group interactions across baseline, post-intervention, and follow-up assessments. Normality assumptions will be examined using distributional metrics, and data transformations will be applied when required to satisfy model assumptions. Machine learning procedures for cluster assignment will be conducted using dedicated data mining software, while inferential statistical analyses will be executed using established statistical packages. Effect sizes will be interpreted alongside statistical significance to contextualize clinical relevance. By integrating multidimensional kinematic profiling with subgroup-specific rehabilitation strategies, this study seeks to determine whether classification-based training improves clinical outcomes, enhances neuromuscular adaptation, and reduces variability in treatment response compared with conventional non-individualized kinematic rehabilitation. In addition, the study aims to clarify mechanistic relationships between improvements in cervical sensorimotor control and changes in pain intensity, disability, dizziness, visual symptoms, and oculomotor function. The findings are expected to contribute to development of personalized, mechanism-based rehabilitation models aimed at reducing recurrence, limiting chronicity, and improving long-term functional outcomes in individuals with idiopathic neck pain.