795 - Wearable Sensor Analysis of Leg Movements Differentiates Risk Status in Infants At Risk for Cerebral Palsy During the First Four Months of Life

Publication Number: 795.7041

Mustafa A. Ghazi, Children's Hospital Los Angeles, Los Angeles, CA, United States; Laura A. Prosser, University of Pennsylvania; Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Thubi H. Kolobe, University of Oklahoma Health Sciences, Sausalito, CA, United States; Andrew H. Fagg, University of Oklahoma, Norman, OK, United States; Julie C. Skorup, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Samuel Pierce, Children’s Hospital of Philadelphia, Philadelphia, PA, United States; Noor Ruwaih, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Beth Smith, Children's Hospital Los Angeles, Los Angeles, CA, United States

Background: We present data from the observational period of an 18-month intervention study focused on locomotor learning (NCT04561232). Infants who experienced a brain injury were observed from 1-4 months of age (corrected). If infants were at high risk for cerebral palsy (CP) at 4 months of age, they progressed to an intervention phase where they received intensive locomotor training. Our focus is on early identification of high risk for CP using wearable sensors.

Objective: To test for group differences in wearable sensor leg movement data collected across 1-4 months of age between infants at high risk for CP at 4 months of age compared to those who were no longer at high risk for CP at 4 months of age.

Design/Methods: We enrolled 60 infants with a history of an early brain injury associated with risk for CP. Starting at 1 month of age, we recorded full-day leg movement data using inertial sensors on each ankle. Data were recorded for 2 consecutive days every month, for 4 months. At 4 months of age, the infants were evaluated to determine whether they were at high risk for CP (n=36 at high risk/HR and n=22 not at high risk/NHR), using General Movements Assessment and Test for Infant Motor Performance. Previously validated kinematic characteristics (average number of movements per hour awake, median movement duration, median acceleration, and fuzzy entropy of peak acceleration) of full-day infant leg movement were derived from tri-axial accelerometer and gyroscope data. We used independent t-tests and Mann-Witney U test to test for group differences in sensor variables between HR and NHR groups at month 1 and in change between months 1 and 4.

Results: For month 1, we observed a significant group difference in number of movements per hour awake (HR:Med=220.8, IQR=199.4; NHR:Med=348.4, IQR=265.1; p=0.002) and movement duration (in seconds, HR:µ=0.262, SD=0.029; NHR:µ=0.287, SD=0.023; p=0.004). There were not group differences in median acceleration or fuzzy entropy. For the amount of change from month 1 to month 4, we did not observe any significant group differences.

Conclusion(s): Number of movements per hour awake and movement duration differentiated between infant risk groups at 1 month of age, supporting the potential for early differentiation of high risk status. Group differences appear to be maintained across months 1-4 as there was not a significant group difference in amount of change between months 1 and 4. Future work will analyze the early movement data in relation to functional and diagnostic outcomes and response to rehabilitation.