2017 Dream Team Projects
Multidimensional Predictors of Major Depressive Disorder and Suicidal Behaviors In Adolescents Ian Gotlib, Trevor Hastie, Holden Maecker, Rachel Manber, Dennis Wall, Jaclyn Schwartz, Tiffany Ho, Erika Manczak
The goal of this longitudinal study is to leverage a well-characterized sample of healthy adolescents who experienced early life stress to integrate multi-system neurobiological and digital phenotypes with machine learning algorithms to identify risk factors and mechanistic targets involved in the onset of depression and engagement in suicidal behaviors. This project will facilitate the development of more timely and precise approaches to the prevention of these debilitating conditions and their devastating consequences.
Our approach is three-pronged:
- Integrate existing and new assessments of neurobiologically (functional and structural MRI, cortisol, immune function) and digitally (sleep, mobility, affective states) derived phenotypes measured prior to the onset of disorder to identify adolescents who are at risk for developing clinically significant depression and engaging in suicidal behaviors
- Use machine learning algorithms suited for multidimensional prediction to create a risk calculator that can be scaled and incorporated into existing health assessments
- Refine the algorithms by applying the risk calculator to data from larger national and international samples of children and young adolescents, allowing us to quantify risk before the first onset of depression and engagement in suicidal ideation
Because we are studying adolescents, we are uniquely positioned to characterize the early emergence of depression and suicidal behaviors that, if undetected, often cascade into lifelong difficulties. Our goal is perfectly aligned with the mission of PHIND to fundamentally revolutionize health care, and specifically, precision mental health.
Precision Diets For Diabetes Prevention Michael Snyder, Manisha Desai, Tracey McLaughlin, Justin Sonnenburg, Christopher Gardner
Predicting Healthy vs. Pathological Aging: Multimodal Biomarkers of Age-Related Memory Change and Risk for Alzheimer's Disease Anthony Wagner, Jennifer McNab, Elizabeth Mormino, Erika Manczak, Carolyn Fredericks, Brian Rutt, Frederick Chin
The Predicting Health in Aging (PHIA) project will address two major precision health goals, leveraging a deeply characterized cohort of 200 healthy older individuals from whom baseline measures of brain structure, brain function, genetics, and CSF biomarkers of risk for Alzheimer’s disease (AD) are collected. PHIA’s first goal is to include a longitudinal follow up assessment that will seek to identify critical biomarkers that predict changes in cognitive and neurobiological health that occur 3.5 years after baseline assessment. PHIA’s second goal is to incorporate additional novel MRI metrics and Tau PET imaging to establish how brain regions known to be impacted early in AD (a) contribute to early changes in memory performance and (b) ultimately predict risk of AD dementia onset over a 7-year window. By using cutting-edge imaging modalities, blood-based and genetic assays, and multivariate analytics to predict the transition from normal to pathological aging, this project will improve the ability to predict long-term health vs. risk of dementia among clinically asymptomatic individuals.
Enabling early cancer detection with lower costs and improved sensitivity from non-invasive genome-wide liquid biopsy tests through novel deep learning analytics and improved chemistry Christina Curtis, George Sledge, Anshul Kundaje, Irene Wapnir, Allison Kurian, Robert West
Develop a novel cfDNA sequencing technology and deep learning analytical framework that achieves state-of-the-art accuracy in distinguishing normal versus pathologic states and tissue-of-origin from clinical samples.
A scalable and robust non-invasive assay to detect cancer early, when it is most curable.