Researchers at Children’s Hospital Boston have discovered a method of identifying infants at high risk for developing autism based on their EEG. The main motivation for this research is identifying children at risk at an age when clinical diagnosis is still impossible, allowing for earlier start of behavioral interventions.
They used machine-learning algorithms to determine the complexity of the resting state EEG signals in a group of 79 infants aged 6 to 24 months, of which 46 had an older sibling with autism. They computed the modified multiscale entropy (mMSE), which indicates the degree of randomness in the signal, a result of the density of neurons, the organization of their connections and the balance of short- and long-distance connections.
Their system had a 80 percent accuracy in identifying high risk infants at nine months of age, with lower accuracy at other timepoints. For nine month old boys accuracy was even 100%, while declining at later intervals. Further studies to confirm the findings and to discriminate patterns between different types of autistic spectrum disorders are planned.
The study was published online in BMC Medicine.
Press release: Using EEGs to diagnose autism spectrum disorders in infants…
Children’s Hospital Boston’s blog: The brain whisperer: Tracking EEG footprints of autism and mental illness…
Study abstract: EEG complexity as a biomarker for autism spectrum disorder risk…