Integrating physiotherapy into social prescribing pathways for chronic pain management

BackgroundChronic pain affects 28 million UK adults. While physiotherapy and social prescribing are recognised as effective, evidence on integrating them remains limited. Digital biomarkers could objectively assess movement quality and pain physiology to support clinical decision-making. ObjectivesTo explore whether simple kinematic and electroencephalography (EEG) biomarkers can discriminate between correct versus incorrect exercise performance and high versus low pain states, and to consider how such tools might streamline social prescribing pathways for chronic pain. MethodsSecondary analyses of two open datasets: (1) KERAAL kinematic dataset: 900 recordings of trunk rotation exercises (21 participants) labelled "correct" or "incorrect" by physiotherapists; (2) EEG dataset: 145 resting-state recordings (100 individuals with chronic pain) categorised as high or low pain. Classification models (logistic regression, support vector machines, gradient boosting) were trained on extracted kinematic features (range of motion, smoothness, symmetry, jerk, coordination) and EEG features (frontal alpha asymmetry, connectivity, sample entropy, aperiodic exponents). Performance was assessed using accuracy, F1 score, AUC and calibration curves. ResultsKinematic classification achieved near-perfect performance: 100% accuracy and AUC = 1.00. Key features were shoulder abduction peak angle, symmetry index and jerk. EEG classification was strong: 93% accuracy, 0.97 AUC. Global theta and alpha connectivity, frontal alpha asymmetry and sample entropy were most informative. ConclusionAutomated kinematic and EEG markers can reliably differentiate correct exercise performance and current pain state. These digital biomarkers could inform physiotherapists and link workers when triaging patients into social prescribing programmes. Integrating digital assessment into social prescribing may reduce unnecessary appointments, enhance self-management and align care with the biopsychosocial model. Further studies should validate the approach in broader populations and examine implementation in real-world social prescribing services.