Smartphone-based digital phenotyping for the characterization of post-operative recovery of patients undergoing surgery for cervical myelopathy

Abstract

Objective: This study aims at investigating the feasibility of using personal smartphones to characterize patients' mobility after surgery for cervical myelopathy. The specific objectives are to: (1) assess differences in GPS/accelerometer mobility features between the first and fifth postoperative 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. Methods: A population of patients with cervical myelopathy undergoing surgical decompression at the Verona University Hospital was enrolled in the 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 postoperative weeks, generalized linear models to assess recovery trajectories, and Pearson's correlation to evaluate relationships between mobility variables and PROMs. Results: Thirteen patients (3 females, 10 males) 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. VAS, mJOA, and ODI were significantly correlated with all five GPS variables, while 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. Conclusions: The 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 suggests that digital phenotyping could complement traditional assessment methods, offering continuous, objective data to support clinical decision-making and personalized patient care.