Computational modeling of psychiatric illnesses via well-defined neurophysiological and neurocognitive biomarkers

A good deal of recent research has centered on the identification of biomarkers and endophenotypic measures of psychiatric illnesses using in vivo and in vitro studies. This is understandable, as these measures-as opposed to complex clinical phenotypes-may be more closely related to neurobiological and genetic vulnerabilities. However, instantiation of such biomarkers using computational models-in silico studies-has received less attention. This approach could become increasingly important, given the wealth of detailed information produced by recent basic neuroscience research, and increasing availability of high capacity computing platforms. The purpose of this review is to survey the current state of the art of research in this area. We discuss computational approaches to schizophrenia, bipolar disorder, Alzheimer's disease, fragile X syndrome and autism, and argue that it represents a promising and underappreciated research modality. In conclusion, we outline specific avenues for future research; also, potential uses of in silico models to conduct "virtual experiments" and to generate novel hypotheses, and as an aid in neuropsychiatric drug development are discussed.