A framework for advancing mechanistic neuro-behavioral biomarkers in psychiatry.

Abstract

Neuropsychiatry has yet to surmount the fundamental challenge of mapping behavioral pathology to its underlying neural pathology. This gap limits the development of treatments for specific circuit pathologies, despite the great potential of neuroimaging measures. We show that the field may be moving towards refining limited statistical frameworks, while clinically translatable solutions could emerge with broader considerations. We posit that the failure to operate within a formalism that defines falsifiable parameters may have hindered progress. Here we propose a provisional formalism with the intention to bring into focus elements that seem necessary to advance the development of precision treatments in psychiatry. Specifically, we propose that this formalism should consider three defining axes: i) type of mechanism; ii) severity of mechanism; and iii) time. These three axes define a 3-dimensional dynamic mechanism complexity space. In turn, we posit that at any point in this mechanism complexity space there are embedded neuro-behavioral sub-spaces or 'geometries' for neural-to-symptom variation, which themselves are multi-dimensional and dynamic. This formalism provides for the mapping of the mechanism complexity space to a neuro-behavioral sub-space. Furthermore, we articulate how this formalism accommodates integration of spectra from genetics and systems biology (transcriptomics, epigenomics, etc.) to neural mechanisms, symptom variance and ultimately taxons of mental illness (i.e. categories). Finally, we argue that this formalism creates an opportunity to evaluate different types of treatment development that map onto dynamically evolving mechanisms of illness that are likely a hallmark feature of neuropsychiatric illness.