318 Classifying Peri-operative Course and Outcomes in Spine Surgery Using Smartphone-Based Accelerometry

Abstract

INTRODUCTION: Spine surgery outcomes are currently assessed using patient-reported outcome measures (ODI, EQ-5D). However, such tools are limited by the inherent subjectivity of the data, as well as collection at discrete timepoints without regard to each patient’s unique clinical course. In contrast, patient mobility data streamed from smartphones with built-in accelerometers offer a more objective and continuous measure of surgical outcome. METHODS: 14 patients who underwent elective lumbar decompressive surgery were included. Patient data were remotely exported from the Apple Health mobile application. A time series analysis was conducted on steps-per-day across a 2-year peri-operative period. Using a data-driven approach, up to five distinct temporal epochs were identified from each patient’s time series: 1) pre-operative baseline; 2) acute pre-operative decline, indicating either an acute event or an acute-on-chronic decline; 3) immediate post-operative recovery; 4) full recovery; 5) secondary decline from fully-recovered state. RESULTS: Mobility data revealed acute pre-operative decline in 10 patients (onset 11.8 ± 2.9 weeks prior to surgery); the remaining 4 patients demonstrated a chronic presentation (no Epoch 2). These presentations were concordant with clinical documentation in 13/14 patients. Following the immediate post-operative recovery period (duration 20.6 ± 4.9 weeks), 11 of 14 patients achieved a full recovery, indicating a return in mobility to levels commensurate to or better than the patient’s pre-operative baseline. Of these, 2 patients subsequently experienced a secondary decline beginning 27.2.1 ± 9.9 weeks after surgery. During the full recovery period, subject-level daily steps improved by 80% ± 33% relative to before surgery (p = 0.002). CONCLUSION: The peri-operative clinical course of spine surgery patients was classified from smartphone-based mobility data using a data-driven method. Our findings highlight the potential of using mobility data for pre-operative counseling, as an objective surgical outcome measure, and for remote patient monitoring beyond the immediate post-surgical period.