Miley, Kathleen2024-03-292024-03-292022-02https://hdl.handle.net/11299/261989University of Minnesota Ph.D. dissertation. February 2022. Major: Nursing. Advisors: Barbara McMorris, Connie White Delaney. 1 computer file (PDF); xiv, 208 pages.Background: Schizophrenia and other psychotic disorders are characterized by severe disability in social functioning, reducing quality of life, increasing risk for poor health outcomes, and causing significant personal and societal burden. Remediating social functioning impairments is an urgent clinical need, however progress has been hindered by a poor understanding of bio-behavioral underpinnings of functional decline, and the resulting lack of both prognostic tools to identify individuals at risk for poor outcomes and robustly effective interventions to promote functional recovery. This dissertation has an overarching purpose to improve the understanding, identification, and remediation of social functioning deficits in schizophrenia spectrum disorders by leveraging data-driven approaches. Three manuscripts are presented. Manuscript 1 critically reviews twelve studies to characterize the state of the science of individual prognostic models for functional outcomes in schizophrenia spectrum disorders. Findings indicate that development of prognostic tools is in an early stage, with a wide range of accuracies, and no clear advantage of utilizing one data modality (i.e., neurobiological data, clinical data, or functional data) over another. Results highlight a need to evaluate and directly compare predictive models which utilize different predictor modalities to understand how to optimally balance accuracy and clinical usability. Manuscript 2 presents a study aimed to develop individual prediction models for social functioning from integrated bio-behavioral data and identify which predictors are most important for social functioning using machine learning. With data from the Human Connectome Project Healthy Young Adult sample (age 22-35, N=1,101) and machine learning methods, four prognostic models were built from variable sets of brain morphology to behavior with increasing complexity: 1) brain-only model, 2) brain-cognition model, 3) cognition-behavioral model, and 4) combined brain-cognition-behavioral model. Results show that the combined brain-cognition-behavioral and cognition-behavioral models significantly predicted social functioning with nearly identical accuracy (R2 =0.53, 95% CI [0.38, 0.62] for each model), whereas the brain-only and brain-cognition models performed significantly worse (R2 = 0.06, 95% CI [-0.07, 0.16] and R2 = 0.11 95% CI [-0.05, 0.23], respectively). Negative affect, psychological wellbeing, extraversion, withdrawal, and cortical thickness of the rostral middle-frontal and superior-temporal brain regions were the most important predictors. These results suggest that prognostic models relying on behavioral data may promote clinical usability while maintaining predictive accuracy, and identify potentially important risk markers to be explored in future research. Manuscript 3 shifts the focus to identifying potential causes of functional outcomes that could be high impact treatment targets in first episode schizophrenia. We used demographic, clinical, and psychosocial measures for 276 participants from the Recovery After an Initial Schizophrenia Episode Early Treatment Program (RAISE-ETP) trial and a causal discovery algorithm, Greedy Fast Causal Inference, to model causal relationships across baseline variables and six-month social and occupational functioning. Results were validated in an independent dataset. Our primary finding was a modeled causal pathway from baseline socio-affective capacity to motivation, and from motivation to both social and occupational functioning at six months. These findings indicate that socio-affective abilities and motivation are specific high-impact treatment needs that must be addressed to promote optimal social and occupational recovery and highlight the need to integrate evidenced based treatments for these areas into gold-standard care models to promote social recovery. Conclusions: This dissertation leverages data-driven approaches to provide foundational knowledge for developing individual prognostic models for social functioning and to guide clinical research seeking to fill critical unmet treatment needs for the remediation of functional impairments. Research and clinical agendas must continue to advance the science toward ensuring that social recovery is the expectation of mental health treatment through early identification of individuals at risk for functional decline and innovative treatments which enhance their functioning. Further synthesis and implications of this work are explored in the concluding chapter.enmachine learningpsychosisrecoveryschizophreniaserious mental illnesssocial functioningUnderstanding Social Functioning Deficits in Health and First Episode Schizophrenia: A Data-driven Approach Towards Improved Identification and TreatmentThesis or Dissertation