Tal Yarkoni - US grants

DESCRIPTION (provided by applicant): Efficient delivery of health care services depends on people's ability to accurately assess the severity of their pain and seek medical care promptly when it is needed. Yet empirical studies have identified many instances in which people systematically misvalue their pain-e.g., patients experiencing heart failure may delay seeking medical attention for several hours after the onset of severe physical symptoms, often with lethal consequences. The goal of the proposed research is to investigate the psychological and neural mechanisms underlying the valuation of pain and explain why and how failures of valuation occur. Behavioral and functional neuroimaging experiments will investigate (a) the relation between different types of valuation of pain (e.g., experienced versus remembered pain, (b) the relation between individual differences in pain valuation and dispositional differences in negative affect, and (c) the neural pathways that mediate the influence of different types of pain valuation on decision-making. By advancing our understanding of the psychological and neural mechanisms involved in pain-related decision-making, the proposed experiments can potentially help to improve measurement of pain in clinical settings and develop novel interventions to enhance the quality of patients'pain-related decisions. Public health relevance: This research seeks to understand how people determine the value of their pain when making pain-related decisions. The proposed experiments have the potential to explain why it is that people often make poor decisions about pain, e.g., failing to seek prompt medical attention when experiencing severe physical symptoms. By shedding light on the mechanisms involved in pain-related decision-making, the research has the potential to improve measurement of pain in clinical settings and to develop more effective policies for identifying and treating individuals at risk for misreporting of physical symptoms.

DESCRIPTION (provided by applicant): The explosive growth of the human neuroimaging literature has led to major advances in understanding of normal and abnormal human brain function, but has also made aggregation and synthesis of neuroimaging findings increasingly difficult. The goal of this project is to develop an automated software platform for large-scale synthesis of human functional neuroimaging studies. Our work builds directly on an existing software platform (NeuroSynth) and involves key extensions and improvements that focus on (i) aggregation, (ii) coding, (iii) synthesis, and (iv) sharing of functional neuroimaging data. In Aim 1, we will use computational linguistics and bioinformatics data mining techniques to develop new algorithms for automatically extracting activation foci and associated metadata from published neuroimaging articles. In Aim 2, we will use topic-modeling techniques such as Latent Dirichlet Analysis in combination with existing cognitive ontologies such as the Cognitive Atlas to develop structured representations of automatically extracted neuroimaging data. In Aim 3, we will improve the meta-analysis and classification capacities of our existing platform by implementing a state-of- the-art hierarchical Bayesian meta-analysis method recently developed by the research team. Finally, in Aim 4, we will develop a state-of-the-art web interface (://neurosynth.org) that supports real-time, in-browser access to the data, results, and tools produced in Aims 1 - 3. Realizing these objectives will introduce powerful new tools for organizing and synthesizing the neuroimaging literature on an unprecedented scale. These tools will be freely and publicly available to anyone with an internet connection, enabling rapid and efficient application to a broad range of clinical and basic research applications. PUBLIC HEALTH RELEVANCE: Functional neuroimaging techniques such as fMRI have opened a new frontier in efforts to investigate and understand the neural mechanisms of normal and abnormal cognition. However, the rapidly expanding scope of the literature makes distillation and synthesis of brain imaging findings increasingly challenging. The goal of this project is to develop a new software platform for automated aggregation, synthesis, and sharing of published neuroimaging results, with the potential to advance understanding of mechanisms underlying mental health disorders.

Project Summary The widespread introduction of functional magnetic resonance imaging (fMRI) over the past two decades has revolutionized the study of human brain and cognitive function in healthy and clinical populations. Yet the potential utility of fMRI remains constrained by its resource-intensive nature. Because even small, exploratory studies are expensive to conduct, biomedical researchers can collectively test only a small fraction of the research hypotheses that are in principle amenable to fMRI investigation. There is an urgent need for novel methodological approaches that enable rapid and efficient testing of novel theoretical hypotheses by reusing existing fMRI datasets rather than acquiring new ones. To help achieve this goal, we propose a new platform called NeuroScout that will support rapid and flexible cloud-based analysis of existing functional fMRI datasets. Our approach differs importantly from previous infrastructure projects in that, rather than developing a domain-general neuroimaging platform, we focus on extracting maximum utility from a limited set of fMRI experiments--namely, those that use intrinsically high dimensional stimuli such as movies and audio narratives. The proposed work encompasses three Specific Aims. Aim 1 focuses on reducing the burden of re-analyzing existing fMRI by automating much of the analysis process and allowing researchers to easily execute their analyses in the cloud. Aim 2 increases analytical flexibility by developing highly extensible tools for multimodal stimulus annotation. Aim 3 focuses on incentivizing platform use by integrating NeuroScout outputs with existing data sharing, visualization and interpretation platforms such as NeuroVault and Neurosynth. When fully deployed, the NeuroScout platform will provide a turnkey solution for extremely rapid analysis and visualization of existing fMRI data at a marginal cost very close to zero. Researchers will be able to iteratively test and refine hypotheses in domains ranging from visual word recognition to social cognition, and interactively visualize and share their results with the broader research community at the push of a button.

PROJECT SUMMARY The human functional neuroimaging literature has experienced explosive growth over the last two decades. In an effort to make sense of this literature, neuroimaging researchers have developed quantitative meta-analysis methods that can aggregate and synthesize the results of hundreds or thousands of studies. In the previous period of this project, we introduced a web-based platform called Neurosynth that supports automated meta- analysis of the fMRI literature at large scale. Neurosynth has become a widely used resource within then neuroimaging community; however, like other meta-analysis approaches to fMRI, it currently supports analysis only of sparse, discrete activations previously reported in published studies. This coordinate-based meta- analysis (CBMA) approach is inferior in many respects to image-based meta-analysis (IBMA) approaches that operate over continuous whole-brain statistical maps. A community-wide shift from CBMA to IBMA would considerably improve sensitivity and specificity, and allow a much broader range of mixed-effects meta- analysis models to be fit to fMRI data. Our overarching goal in the present project period is to contribute to such a shift by extending the existing Neurosynth platform into a turnkey solution for image-based meta- analysis. In Aim 1, we will create a centralized database of whole-brain statistical maps, providing a rich data source for large-scale image-based meta-analyses. In Aim 2, we will add new web-based interfaces to Neurosynth that enable users to easily (i) edit, validate, and annotate data from individual studies, and (ii) organize data from hundreds or thousands of studies into sophisticated image-based meta-analyses that can be readily executed on local or cloud computing resources. In Aim 3, we will develop a reference open-source software package (PyCIBMA) for efficient mixed-effects meta-analysis of fMRI data, providing the community with a uniform interface for fMRI meta-analysis that complies with current open standards and specifications. Realizing these objectives will introduce powerful new tools for synthesizing the neuroimaging literature at a large scale and with unprecedented resolution. These tools will be freely and publicly available to anyone with an internet connection, enabling rapid and efficient application to a broad range of clinical and basic research applications.