Interesting paper which uses machine learning techniques (bagged-multiview clustering) to analyse the results of MRI, IQ (Wechsler or Stanford-Binet), a scale of autism symptoms (SCQ), a scale of inattention symptoms (SWAN), and a scale of obsessive-compulsive traits (TOCS), across four groups: those with a confirmed primary diagnosis of ASD, those with a confirmed primary diagnosis of ADHD, those with a confirmed primary diagnosis of OCD, and controls without any neurodevelopmental/psychiatric/neurological diagnosis.
The clusters they derived had weak alignment with the diagnostic labels, from which they concluded that the diagnostic labels have poor biological validity.
(Which is not really that novel a conclusion – the belief that DSM-5 diagnostic labels lack biological validity is very widespread in the research community; the methodology of the paper is arguably more novel than its conclusion.)
The clusters they derived had weak alignment with the diagnostic labels, from which they concluded that the diagnostic labels have poor biological validity.
(Which is not really that novel a conclusion – the belief that DSM-5 diagnostic labels lack biological validity is very widespread in the research community; the methodology of the paper is arguably more novel than its conclusion.)