Mental disorders represent immense healthcare burdens. Intense efforts and resources have been devoted to developing pharmacological and behavioral treatments for mental disorders, but no universally effective treatments are available. Considerable heterogeneity exists in treatment response among individuals with mental disorders, in part because an individual patient's psychosocial characteristics and/or biomarkers are not accounted for when selecting among available treatment options. Barriers to implement personalized treatments in clinical psychiatry include a lack of evidence-based, clinically interpretable, individualized treatment rules (ITRs), a lack of power to detect treatment modifiers from a single study, and a lack of reproducibility for treatment rules estimated from single studies. We propose analytic solutions to tackle these barriers. Specifically, we will provide integrative machine learning methods to build powerful, yet interpretable, individualized treatment strategies that can be easily applied in clinical practice. We will integrate evidence of ITRs identified in multiple randomized controlled trials (RCTs) to increase robustness and reproducibility.
In Aim 1, we will provide piece-wise linear decision trees that are transparent, interpretable, and that have guaranteed performance. Our decision trees will simultaneously identify the optimal treatment for a given patient (qualitative interaction) and subgroups of patients with large benefit (quantitative interaction).
In Aim 2, we propose a novel integrative analysis to synthesize evidence across trials and provide an integrative ITR that improves efficiency and reproducibility. Our method does not require all studies to collect common sets of variables and thus allows evidence to be combined from ITRs identified in recent RCTs that collected emerging biomarkers (e.g., neuroimaging measures) with earlier RCTs that focused on clinical and behavioral markers. In response to the National Institute of Mental Health Strategic Plan on Research Domain Criteria (RDoC) to center mental health research around ?dimensional psychological constructs? shared across disorders, the methods will be applied to a wide range of RCTs that recruited patients with major depressive disorder and other co-morbid mental disorders. This strategy allows examination of ITRs for constructs shared across disorders and will increase generalizability. We will apply our methods to various RCTs, including data available from the National Database for Clinical Trials Related to Mental Illness. This research will bridge approaches for personalized medicine and integrative analysis in an effort to better understand the complex interplay between biomarkers and clinical manifestations in the context of selecting the best treatments for patients with mental disorders.
Treatment responses for mental disorders are inadequate and considerable heterogeneity is observed, in part because an individual patient's clinical, psychosocial, and/or biological markers are not accounted for when select- ing treatments among available options. This research proposes novel analytic methods to discover new powerful, yet interpretable personalized treatment strategies and integrate evidence of strategies identified in multiple prior studies to increase robustness and reproducibility.
