American researchers asked a family group to interact with their children with objects and people. They tried it Eight model of automated learning To diagnose autism, it allows "streamlining processes and making them much more efficient", according to the research published in the scientific journal PLOS Medicine.
The study was developed by a group of the School of Medicine of the University of Stanford and led by Professor Dennis Walle, of the California Pediatric and Scientific Scientific Data Biomedical Institute of California.
Each model "involves a set of algorithms" Characteristics of children between 5 and 12 years old and the overall score of childhood autism, "he explained.
How videos were treated
Walle said to evaluate models, families were asked to submit home video exams for five minutes. The faces and the hands of the children appeared and they captured "social interaction and the use of toys, pencils and tools".. Of these, 116 boys were diagnosed with autism in the age range of 4 years and 10 months and were developing another 46 (two and 11 months).
Expert reviews of nine review videos 30 questions questionnaire With "yes" or "no" responses, according to the typical behavior of autism, eight mathematical models were introduced.
The best models for the best results identified 94.5% of autistic children and 77.4% without children with autism. Evaluated results to verify 66 other videosHalf of the children with autism. Children with self-determination identified 87.8% of the same model, and 72.7% of people with no disorder.
Another advantage of using domestic videos diagnosis "Children are taken in their natural environment", "unlike rheumatism and artificial arterial behavior that can lead to atypical moderation". "We have identified a small group of behavioral characteristics that are very aligned with clinical results, and experts do not rate these features quickly and independently in a virtual environment, in one minute," said Walle.