Diagnosing Autism Spectrum Disorder (ASD) in adulthood is often a complex and lengthy process. A key challenge in this field is the need to differentiate between Autism and other conditions also characterized by difficulties in social interaction, such as Borderline Personality Disorder (BPD). Currently, there is a lack of sufficiently valid and accessible tools for this differential diagnosis, which can lead to delays and difficulties in receiving appropriate support and treatment.
An innovative study, published in the BMJ Ment Health journal on July 30, 2026, examined the feasibility of using advanced technologies for differential diagnosis. The researchers' goal was to develop a model that could distinguish between autistic people and people with Borderline Personality Disorder, thereby streamlining the diagnostic process.
For the study, rich data was collected from reciprocal interactions - two-sided conversations. The researchers analyzed over 800 different features from these conversations, focusing on core areas of social interaction relevant to both populations. Among the features examined: speech patterns, facial expressions, amount of movement, and interpersonal synchronization in facial expressions.
The study findings indicate that initial models focusing on features such as facial expressions during speech and listening, speech patterns, synchronization in facial expressions, and amount of movement - all succeeded in distinguishing between autistic people and people with Borderline Personality Disorder to a degree greater than chance. Combining all these models further increased accuracy, with the presented algorithm achieving a balanced accuracy level of almost 82%, all based on just 20 minutes of conversation.
This initial feasibility study ("proof-of-concept") reveals immense potential for the use of advanced diagnostic technologies. The researchers report that data collection only requires microphones and webcams, and feature extraction is performed automatically, making the approach highly objective, scalable, and user-friendly.
For autistic adults in Israel, such a development could be important news. It offers the possibility of significantly streamlining the differential diagnostic process, which could lead to a faster and more accurate diagnosis. A quicker diagnosis can reduce the distress of those waiting for assessment and shorten the waiting time for a personalized treatment and support plan.
It should be noted that this is an initial feasibility study ("proof-of-concept"), and therefore further clinical validation is required to confirm its findings and integrate the algorithm into routine clinical diagnosis in the future.