Family video can help diagnose autism, according to a study

A new discovery could facilitate the diagnosis of autism, which is currently complex and time consuming. A study published in Plos Medicine may have revealed a faster and more effective way to diagnose the disorder through home videos that show children interacting in familiar contexts. Indeed, according to researchers, the doctor's office causes atypical behavior in children, which can interfere in the identification of the disorder. Even observed in places where they are accustomed, even if it is not present, the diagnosis can be made more easily because the behaviors presented are usually natural.

This discovery is extremely important because previous research has revealed that behavioral therapies for autism work best when they start before the age of 5 years. However, because of the difficulty of the doctors to identify the necessary traits to confirm the diagnosis, in addition to the fact that some families have to deal with long queues to be referred to a specialist, the treatment may be deferred . For researchers, the use of videos can speed up the diagnostic process with a precision guarantee of up to 80%, allowing treatment to begin as soon as possible.

In addition, the team revealed that by using fairly simple mechanisms, non-specialist pediatricians could use videos during consultations to observe unusual behaviors; these observations would be placed on the map for surveillance. If the child has enough features of the autism spectrum (ASD), the doctor may feel confident to make the diagnosis or refer the patient to a specialist who can confirm the suspicion.

Method of observation
The first step in diagnosing autism is to identify the symptoms. To do this, the team at Stanford University in the United States used a machine to determine what characteristics of children's behavior should be classified as autism.

To reduce the list of symptoms, scientists used computers to define five to twelve most relevant traits for diagnosis. They also created an algorithm capable of establishing a score for each characteristic of autism; these points would be added to each identified trait and at the end, the child would have a score that would approach or distance him from the diagnosis of ASD, in addition to helping to rank the severity of the trouble.

Recruitment

Through social networks, researchers recruited families with autistic children without trouble and asked them to send videos of their children home. The screening examined criteria such as showing the child's face and hands, showing a direct social interaction or videos in which he was playing with objects. The team received 162 videos (lasting 1 to 5 minutes). Of these, 116 were children with autism and 46 children with normal development. The average age of the first group was 4 years and 10 months; in the second, 2 years and 11 months.

appraisers

The researchers selected nine critics to view the videos and identified specific behaviors through questionnaires with 30 objective questions (yes or no) to diagnose autism. Each of them has received simple instructions on how the assessment should be carried out, taking into account behaviors such as expressive language (verbal or non-verbal), the expression of emotions, eye contact and the attention of objects.

After each participant evaluated 50 videos, the scientists determined that three reviews were the minimum number needed to generate a reliable score. The remaining videos were distributed randomly to reviewers, so at least three of them analyzed the same video.

Fast and efficient diagnosis
According to the researchers, each evaluator took an average of four minutes to observe and identify behaviors. The accuracy of identifying the presence or absence of autism was 88.9%, with correct labeling of 94.5% of children with autism and 77.4% of children with autism. non-autistic children. To confirm the results, the researchers repeated the experiment with 66 additional videos – 33 autistic children and 33 non-autistic children. The correct identification reached 87.8% for autism and 72.7% for normal development.

"We have shown that it is possible to identify a small set of behavioral characteristics with high clinical outcomes, not even specialists [em autismo] can perform the assessment in an online environment in minutes, "said Dennis Wall, lead author of the study, in a note.For him, this new diagnostic model could help identify autism even at home. the youngest children – the age range in which the diagnosis is even more complex.By being fast, the feature could even be used during routine queries.

While this represents hope for families, the researchers point out that more studies are needed to confirm the effectiveness of the method in other populations.