Supplementary file 1_Smartphone-based digital phenotyping for characterizing post-operative recovery in patients undergoing surgery for cervical myelopathy.docx

ObjectiveThis study aimed at investigating the feasibility of using personal smartphones to characterize mobility in patients after surgery for cervical myelopathy. The specific objectives are to (1) assess differences in global positioning system (GPS)/accelerometer mobility features between the first and fifth post-operative weeks, (2) evaluate differences in recovery trajectories between anterior and posterior surgical approaches, and (3) analyze correlations between pain/disability scores and passively acquired GPS/accelerometer mobility variables.
MethodsA population of patients with cervical myelopathy undergoing surgical decompression at the Verona University Hospital was enrolled in this study. Data collection included passively acquired GPS and accelerometer data from personal smartphones, patient-reported outcome measures (PROMs), and demographic/surgical information. Statistical analysis included descriptive statistics, Wilcoxon rank sum tests for comparing mobility variables between the first and fifth post-operative weeks, generalized linear models to assess recovery trajectories, and the Pearson’s correlation to evaluate relationships between mobility variables and PROMs.
ResultsA total of 13 patients (3 female and 10 male patients) were included. All five mobility variables showed significant improvement from post-operative week 1 to week 5 (p < 0.001). The surgical approach was significantly associated with recovery trajectories for Home Duration, Steps, and Significant Location Count. Patients who underwent anterior surgery generally showed higher initial mobility levels and steeper recovery trajectories. The visual analog scale (VAS), the modified Japanese Orthopedic Association (mJOA), and the Oswestry Disability Index (ODI) were significantly correlated with all five GPS variables, whereas the Neck Disability Index (NDI) was only significantly correlated with Distance Traveled and Steps. Linear relationships between mobility variables and PROMs were observed, with increasing uncertainty at higher mobility levels.
ConclusionThe study demonstrates the potential of smartphone-derived mobility data as a valuable tool for characterizing post-operative recovery in cervical myelopathy patients. The ability to characterize recovery trajectories and correlations with established clinical measures indicates that digital phenotyping could complement traditional assessment methods, offering continuous, objective data to support clinical decision-making and personalized patient care.