Project Summary/Abstract The lack of accessible, individual-level measurements suited to inform clinical decision-making is a critically unmet need in depression, the leading cause of disability globally. Persistent cognitive impairments from depression are a major contributor to disability. Our objective is to optimize, validate, and deploy a clinical cognitive signature using behavioral measures that have a basis in neural mechanisms, enabling individualized assessment at scale suited to personalized clinical prognostic and treatment selection decisions. We will extend our pioneering work in identifying a cognitive phenotype of depression derived from computerized behavioral ‘WebNeuro’ tasks that align with the RDoC cognitive control construct, to be complemented by a novel, research- based smartphone ‘BiAffect’ application for finer-grained, passively sampled behavioral metrics. In Aim 1, we will optimize a clinical cognitive signature for individual-level predictions based on our already identified cognitive phenotype. We leverage our unique, large existing multi-modal datasets with common cognitive data elements totaling 3,082 diverse participants. These datasets span participants with major depressive disorder assessed pre-post treatment with pharmacotherapy and behavioral therapy, pre-post naturalistic trajectories, and matched healthy participants from the same sites. We will systematically optimize a clinical cognitive signature by generating trial-by-trial individualized scores on cognitive control tasks, with refined norms, and evaluate these scores in predictive models. We will also refine the mechanistic understanding of the clinical cognitive signature in the subset of participants who also have neuroimaging data. In Aim 2, we will evaluate the clinical cognitive signature in combination with digital phenotyping at scale in a new prospective sample of 1,200 adults with depression, to be recruited remotely and representative of racially/ethnically and socioeconomically diverse population. We will complement WebNeuro with the BiAffect technology, both suited for remote administration, to quantify finer-grained individual variations in behavior throughout the day. This new cohort will complete repeat assessments for symptom and disability outcome predictions over 8 weeks with a 6-month follow-up. In Aim 3, we will validate the clinical cognitive signature for prospective stratification in a randomized clinical trial with 160 participants from the Stanford Bay Area and Chicago sites. We will prospectively identify participants with a prominent clinical cognitive signature (designated as C+) and those with a relative absence of the signature (designated as C-). Participants will be randomly assigned to receive sertraline plus guanfacine or sertraline plus placebo. Guanfacine is chosen because it has been shown to ameliorate cognitive deficits in depression based on the published preliminary findings from our team. The expected end product will be a clinically validated cognitive signature using a behavioral assessment tool that can be readily scaled and translated into routine clinical practice to inform prognostic and tailored treatment decisions. The project will generate a unique FAIR- compliant dataset enabling future scaling using machine learning and AI.

Project Narrative A major obstacle to advancing precision psychiatry is the lack of individual-level measures suited to clinical decisions. We will optimize and validate a clinical signature measuring a cognitive phenotype linked to treatment non-response and disability in depression, enabling individualized prediction of risk and treatment selection.