Development and validation of an Arabic language eye-tracking paradigm for the early screening and diagnosis of autism spectrum disorders in Qatar.

The developmental-delay group had only 12 children, and all controls were analysed together, so the study does not show how the index separates ASD from other developmental delays. The control group included 39% siblings of children with ASD, which may make controls less typical of the general population. The Arabic index was scored using the original English study's mean and standard deviation rather than local Qatari norms. The higher machine-learning accuracies were exploratory, used the same indicators, and were not replicated in a new sample. Restricting to cases with better eye-tracking data did not improve accuracy (AUC=.689), likely because more affected ASD cases were lost. ADOS-2 severity scores were available only for the ASD group and had a restricted range (82.6% between 3 and 8), so the lack of correlation with the index may reflect limited score spread. The retest sample had significantly higher proportions of language/communication disorders and global developmental delay/intellectual disability than the baseline sample, so retest findings may not represent the full sample. IQ results were missing for some participants. The study validated a laboratory eye-tracking measure; it did not test the tool in routine clinical workflow or compare it directly with existing tools in practice.

The supplied metadata lists the paper as Research Support, Non-U.S. Gov't. No author conflict-of-interest statement or specific funder is given in the supplied text.

Do not treat the Arabic eye-tracking index as a stand-alone diagnostic test. It showed good differentiation of ASD from controls (AUC=.730), but it did not correlate with ADOS-2 severity, and the higher machine-learning accuracies (.799 and .768) were exploratory and need replication.