PROJECT SUMMARY
Biologically-grounded data-driven stratification of disease has revolutionized most of modern medicine. Yet,
psychiatric disorders—a major public health concern—are still defined strictly on the basis of subjective signs
and symptoms. Newly available large-scale human neurobiological data offer an unprecedented opportunity for
biologically-grounded data-driven stratification of psychiatric disorders. However, conventional data analytical
procedures are ill-suited for such investigations, and new computational approaches are needed. The goal of
this proposal is to develop and validate a novel computational framework for identifying, validating, and
characterizing biological subtypes of psychiatric disorders, and to apply the framework to the study of autism.
We aim to leverage newly available large-scale open-source human brain imaging, phenotypic and
transcriptomic data from consortia and repositories around the world, as well as data we have acquired at
Stanford University, combined with our recent work in brain circuit analysis and modeling methods and
advances in machine learning. To achieve these goals, we propose three aims. In Aim 1, we will develop and
validate novel computational methods to extract individual-level neural fingerprint, in particular brain circuit
dynamics, using human brain imaging data. In Aim 2, we will develop analytical procedures to stratify
psychiatric disorders using neural fingerprints and apply the procedures to identify and validate biological
subtypes of autism. In Aim 3, we will perform integrative analysis to determine the transcriptomic signatures of
the identified biological subtypes, using neural fingerprints and human brain gene expression data. The
proposed work is highly relevant to the strategic plan of the NLM to accelerate discovery and advance health
through data-driven research. Through the successful completion of the work described here, the proposed
studies will add new knowledge to our current understanding of the etiology of autism and, crucially, provide a
new computational framework for improved stratification of other psychiatric disorders. Ultimately, these
advances will lead to better diagnosis and more effective treatments for psychiatric disorders and, more
broadly, advance precision medicine.
