Use of QEEG-based cluster analysis for subtyping within psychiatric diagnoses
Methods have been developed to predict the differential responses of subtypes of patients with the same DSM diagnosis, using patterns of QEEG and ERP features. Discriminant functions were constructed that reliably separated normal subjects from patients suffering from a variety of brain dysfunctions. Mathematical cluster analysis methods have been developed to sub-divide heterogeneous groups of patients, with the same DSM-3 or DSM-4 clinical diagnosis. By cluster analysis based on multivariate patterns of QEEG features, large cohorts of patients were parsed into homogeneous subtypes comprised of patients with distinctive profiles of QEEG and ERP abnormalities.
Successful classification methods have been developed to identify patients with various diagnoses who are most likely to display good or poor responses to particular treatments or to display different evolution or outcomes of illness These include: unipolar versus bipolar affective disorders, ADHD responders or non-responders to stimulant therapy, crack cocaine addicts who will or will not respond to rehabilitation treatment, OCD responders or non-responders to SSRI medications, and elderly patients who are objective asymptomatic and will either remain cognitively normal or deteriorate to mild or severe dementia. Subtyping of large cohorts of patients in psychotic states has revealed multiple subtypes each containing patients diagnosed as Schizophrenic or Major Affective Disorder, with preliminary indication of possible differential responses to treatment.
Recently, particular interest has been focused on follow-up studies of the long term effects of crack cocaine on brain functions, as assessed by QEEG and ERPs.