Diagnose autism by analyzing finger movements? This is the hypothesis put forward by Canadian and Israeli researchers. Two French experts react to this technique which they consider reducing and insufficient.
Detecting autism by simply analyzing the movement of the fingers is the promise of a team of researchers from the York University psychology department in Toronto (Canada) and the University of Tel Aviv (Israel). According to them, artificial intelligence (AI) would make it possible to detect “Motor alterations”, One of the key characteristics of autism spectrum disorders (TSA). Objective ? Promote early diagnosis. According to scientists, these disorders “Could indeed be recognized in the first year of age, before the appearance of classic social communication deficits”. Their work has been published in the International Review Autism Research April 17, 2025.
Thirteen cinematic parameters analyzed
To verify their hypothesis, they equipped 59 young adults, including 30 autistic diagnosed, with two motion sensors, on the go and the index. Their mission: to seize, exclusively with these two fingers, five rectangular objects placed 30 centimeters away, on a table. The researchers then analyzed different parameters such as the trajectory of the two fingers, the speed of the opening of gripping, the location, the timing, the stimulation characteristics … These data were then processed by five classification algorithms, trained to distinguish autistic profiles from those who are not.
A high success rate
Verdict ? “We have revealed that the sensitivity -the proportion of properly identified autistic individuals -varied from around 85 % to 89 %, while the specificity -the proportion of properly classified non -autistic individuals -varied between 84 % and 90 %”, The researchers welcome. “These results suggest that the subtle differences in engine control can be effectively captured, offering a promising approach to develop accessible and reliable diagnostic tools for autism”, they believe.
A distrust of “Miracle effects in studies”
In France, experts issue some reservations … Etienne Pot, interministerial delegate for the National Strategy for Neurodevelopment Disorders (TND), calls for “Look at this hypothesis, like many others in the field of autism, with a scientific eye”. “And, given the low staff of the study, we should go much further”, he believes. “I am wary of miracle effects in studies, we have been experiencing a real craze for the Eye Tracking for several years (Editor’s note, which promises to diagnose autism by analyzing the eye movements), However, it has not become a diagnostic tool. »»
Signs “Insufficient” To establish a diagnosis
“The variations in fine motor skills are, certainly, signs but it is not enough to establish a diagnosis, which is mainly based on neurological functioning”, ABONDE CHAMS-DDINE BELKHAYAT, President of the BLUE NETWORK ASSOCIATION AND DAPA of a concerned child. “Autism is a disorder of neurode development, it is not a psychomotor or motor skills disorder”, recalls the co-author of the book Autism: Understand to better accompany. These symptoms must therefore “Being coupled with observation criteria on cognitive functioning and information processing, since autism, before being atypical gestures, stereotypia, routines, etc., it is above all a processing of different information”.
Autism reduced to a motor signature?
Moreover, “Some autistic people have no or very little gestural stereotypes or gripping difficulties so the risk of relying on this single criterion is to miss them or to assign TSA to people who are not”, SUPPUITE CHAMS-DDINE BELKHAYAT. He notably takes the example of non -autistic children who, all small, use flapping (fast beat of the hands), in order to regulate their emotions, “But that disappears over time”. He fears that this type of study will reduce autism to a motor or algorithmic signature, forgetting its diversity.
A diagnostic strategy with several components
“These works nevertheless show that in the future, the diagnosis of autism could call on several tests in different dimensions”, Indicates Etienne Pot, who “Creates much more in a diagnostic strategy with several components than a unique strategy”. Moreover, “All these new tests must be confronted with a reality of very broad autism, from the person with an intellectual impairment to that without TDI, etc … without forgetting the epilepsy which can also influence these tests”, he underlines.
Motor difficulties insufficiently taken into account
However, specialists agree that the motor difficulties of autistic people are not “Most certainly” not sufficiently considered when making a diagnosis. “In general, associated disorders and known comorbidities, both in the somatic and mental health are insufficiently taken into account; Even if we are progressing thanks to our five centers of excellence and all the autism resource centers (CRA) ”, Complete Etienne Pot.
AI, promising but not autonomous tool
The latter said to himself “Very attentive to all developments that could reduce the time of diagnosis”. In this context, AI appears as a “A tool for the future on which I encourage all researchers in our country to watch”. “This technology is indeed very promising, provided they coupled it to human intelligence”, Rebounds Mr. Belkhayat. He specifies: “AI should not be an autonomous tool because the observation of the main autistic characteristics – communication, social interactions, stereotypia, restricted interests – require human knowledge of all that is autism. »» More, “Today, the major issue is not so much the early diagnosis in very young people as early identification, which is everyone’s business, parents but also professionals in the childhood and early childhood sector”, And actually requires increased attention … of the human, concludes the inter -ministerial delegate.
© Pixabay Geralt
“All reproductive and representation rights reserved. © Handicap.fr. This article was written by Cassandre Rogeret, journalist Handicap.fr”
