What is Computational Psychiatry Good For?

Abstract

It is rare in the field of biological psychiatry for hypotheses to be definitively refuted. Rather, topics of investigation drift into and out of fashion, often driven by the initial excitement of technological innovation, followed by the necessary corrective of nuanced or underwhelming clinical results. A well-known example of this is the association between depression and abnormal function of the hypothalamic-pituitary-adrenal axis, as measured using the dexamethasone suppression test (DSST) ( 1 Carroll B.J. Martin F.I. Davies B. Resistance to suppression by dexamethasone of plasma 11-O.H.C.S. levels in severe depressive illness. Br Med J. 1968; 3: 285-287 Crossref PubMed Scopus (219) Google Scholar ). This observation led to a great deal of work investigating whether the association might help us identify useful subtypes of depression ( 2 Schatzberg A.F. Rothschild A.J. Stahl J.B. Bond T.C. Rosenbaum A.H. Lofgren S.B. et al. The dexamethasone suppression test: Identification of subtypes of depression. Am J Psychiatry. 1983; 140: 88-91 Crossref PubMed Scopus (124) Google Scholar ) or help us predict treatment response ( 3 Rush A.J. Weissenburger J. Vasavada N. Orsulak P.J. Fairchild C.J. Dexamethasone suppression test status does not predict differential response to nortriptyline versus amitriptyline. J Clin Psychopharmacol. 1988; 8: 421-425 Crossref PubMed Scopus (1) Google Scholar ). As it turned out, the specificity and predictive value of the DSST was not thought to be of a level that would be useful clinically and the topic has gradually moved out of the spotlight. We are left in the familiar position of knowing that nonsuppression of the DSST is associated with depression, but not being sure how we can use this knowledge to help patients.