Computer-based analysis of general movements reveals stereotypies predicting cerebral palsy

Abstract

Aim: To evaluate a kinematic paradigm of automatic general movements analysis in comparison to clinical assessment in 3-month-old infants and its prediction for neurodevelopmental outcome. Method: Preterm infants at high risk (n=49; 26 males, 23 females) and term infants at low risk (n=18; eight males, 10 females) of developmental impairment were recruited from hospitals around Heidelberg, Germany. Kinematic analysis of general movements by magnet tracking and clinical video-based assessment of general movements were performed at 3 months of age. Neurodevelopmental outcome was evaluated at 2 years. By comparing the general movements of small samples of children with and without cerebral palsy (CP), we developed a kinematic paradigm typical for infants at risk of developing CP. We tested the validity of this paradigm as a tool to predict CP and neurodevelopmental impairment. Results: Clinical assessment correctly identified almost all infants with neurodevelopmental impairment including CP, but did not predict if the infant would be affected by CP or not. The kinematic analysis, in particular the stereotypy score of arm movements, was an excellent predictor of CP, whereas stereotyped repetitive movements of the legs predicted any neurodevelopmental impairment. Interpretation: The automatic assessment of the stereotypy score by magnet tracking in 3-month-old spontaneously moving infants at high risk of developmental abnormalities allowed a valid detection of infants affected and unaffected by CP.