Smartphone-Based Activity Research: Methodology and Key Insights

Ryan W Turlip¹Daksh Chauhan¹Hasan S Ahmad¹Mert Marcel Dagli¹Bonnie Y Hu¹Richard J Chung²Yohannes Ghenbot¹Ben J Gu¹Nisarg Patel³Richelle J Kim⁴Julia Kincaid¹Akash Verma¹Jang W Yoon^1*

• ¹Department of Neurosurgery, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, United States
• ²School of Medicine, University of Virginia, Charlottesville, Virginia, United States
• ³Lewis Katz School of Medicine, Temple University, Philadelphia, Pennsylvania, United States
• ⁴Columbia College, Columbia University, New York City, New York, United States

Background and ObjectivesObjectively studying patient outcomes following surgery has been an important aspect of evidence-based medicine. The current gold-standard – patient reported outcomes measures – provides valuable information but have subjective biases. Smartphones, which passively collect data on physical activity such as daily steps, may provide objective and valuable insight into patient recovery and functional status. This study aims to provide a methodological guide for data collection and analysis of smartphone accelerometer data to assess clinical outcomes following surgery.MethodsPatient health metrics – namely daily steps, distance travelled, and flights climbed – were extracted from patient smartphones using easy-to-download applications. These applications upload the data that smartphone accelerometers passively collect daily to a HIPAA compliant encrypted server while de-identifying the patient’s personal health information. Patients were consented in multiple settings – synchronously during clinical visits or asynchronously over the phone – and could be enrolled during the initial pre-operative visit or well after the surgery.With the patient data acquired, the peri-operative window of selection is determined based on the needs to the study. The timeseries data is then statistically normalized to account for individual baselines and smoothened over a 14-day moving average to minimize noise. Mathematical analysis can be harnessed to study quantifiable recovery and decline periods, which provide continuous and nuanced insight into patient’s health throughout their spine disease and treatment course. Additionally, integrating clinical variables permits computational machine models capable of predicting patient trajectories and guiding clinical decisioning.ConclusionSmartphones offer a new metric for studying patient well-being and outcomes after surgery. The research with them is in its nascent stages but further studies can potentially revolutionize our understanding of spinal disease.