Neuromotor risk probability from automated analysis of crib videos

Spontaneous motor movements, without the presence of external triggers, occur in infants. These general movements, especially those that are called 'Fidgety Movements,' are highly accurate [Einspieler 2016] in predicting neuromotor conditions such as cerebral palsy. Preterm infants are at higher risk for cerebral palsy than full-term infants.

The work by Chambers2019 showed that the movements of infants at risk for neuromotor conditions are considerably different from movements of typically developing children. To their work, that used 85 videos of typically developing children, we added 39 more videos obtained from our own study.

The EMMA AI tool analyzes the crib video, extracts the motor movements and compares it to this 'reference' population of 124 infant videos. The resulting metric is a probability that indicates risk. 

The higher this risk metric is, higher the likelihood that the given infant's movements resemble those of the reference population.

References

Einspieler C, Peharz R, Marschik P. Fidgety movements -- tiny in appearance, but huge in impact. J Pediatr (Rio J). 2016: 92(3 suppl 1): 564-570.

Chambers C, Seethapathi N, Saluja R, Loeb H, Pierce S, Bogen D, Prosser L, Johnson MJ, Kording KP. Computer vision to automatically assess infant neuromotor risk. BioRxiv 2019, 756262. Available online: https://www.biorxiv.org/content/10.1101/756262v1