| Sumario: | Autism spectrum disorder (ASD) is a common neurodevelopmental condition whose diagnosis is often delayed. The feasibility of OlaMind, a new digital system to screen for ASD in verbal children and adolescents without intellectual disability, was tested. Parents in Israel used the instrument concerning their 4-18-year-old children with a doctor-given ASD diagnosis (ASD group: n = 33; 7 females) or without one (Control: n = 44; 20 females). A high proportion of the control group had parent-reported diagnoses of non-ASD conditions whose symptoms may resemble those of ASD (Other: n = 24; 9 females). Those who did not have such diagnoses were considered typically developing (TD: n = 20; 11 females). The algorithm classified 32 (97%) of ASD cases as exhibiting an ASD symptom pattern across DSM-5 clinical diagnostic domains and one 18-year-old male as exhibiting a non-ASD symptom pattern. The algorithm classified 75% of TD cases and 50% of Other cases as non-ASD. There was good correlation between the analytic groups and algorithm classifications. Two Other cases and one TD case were subsequently diagnosed with ASD or with multiple other conditions, respectively. Classification accuracy, sensitiveity, and specificity were 81%, 97%, and 66%, respectively in the full sample, improving to 93%, 97%, and 84%, respectively for the ASD and TD groups after accounting for subsequent known diagnoses. The results demonstrate the feasibility of the OlaMind algorithm. A full validation and reliability study is planned.
|
|---|