Three driving biological projects (DBPs) guide algorithm, tool, and infrastructure development. Each project serves as a testbed for the design, implementation, and validation of tools and services. These DBPs have been selected to illustrate a spectrum of computational algorithms, methods, and tools. They include translational research on molecular profiling of an uncommon disease (pictured at right), sensor networks for personal monitoring, and tools for population-based studies. The iDASH center generates and refines open-source software and infrastructure to support:
Molecular Phenotyping of Kawasaki Disease
Kawasaki Disease, a self-limited acute vasculitis, is the most common cause of acquired heart disease in children. Although coronary artery aneurysms can be prevented by timely administration of high dose intravenous immune globulin (IVIG), the cause is unknown, and researchers suspect an infectious agent that triggers an immunologic reaction in genetically susceptible hosts. In this DBP, we will create molecular phenotypes of patients across the universes of genotype, microRNA, and gene expression by leveraging a repository of RNA and DNA (collected through NIH support: HL69413). We will relate these profiles to patient demographic and clinical data, allowing us to understand how molecular phenotype relates to clinical phenotype and how they may help predict susceptibility, response to treatment, and risk for cardiovascular sequelae.
Post-Marketing Surveillance of Hematologic Medications
Detecting rare events in relatively small collections of data, such as complications of new medications in elderly or minority cohorts, is a difficult problem to address. One domain that can potentially alleviate this problem and thus save lives is post-marketing surveillance of new medications. In this DBP, we will study adverse events associated with four different oral hematologic medications (prasugrel, clopidogrel, warfarin, and dabigatran). The most common bleeding complications will be monitored as well as rare but potentially life-threatening events, including hemorrhagic stroke and thrombolytic thrombocytopenic purpura. This DBP will be a collaborative effort between three institutions: Brigham and Women’s Hospital, Vanderbilt University, and UCSD. Data will be obtained from medical records via research data warehouses and will be analyzed at each institution using several statistical process control and machine learning algorithms.
Individualized Intervention to Enhance Physical Activity
The lack of physical activity (PA) and continual sedentary behavior (SB) over time can contribute to a variety of health problems including obesity and premature morbidity and mortality. Although knowledge about health behavior at the individual and population level has expanded, the integration of these findings into new forms of health behavior interventions is needed. In this DBP, we will create an intervention system to provide individualized feedback to increase PA and decrease SB. We will use wireless accelerometer sensors to measure body movement and position (e.g., seated, standing, walking, running, lying down). Machine learning pattern recognition algorithms will identify and learn a person’s pattern of PA and SB during the day. By identifying patterns, the system will offer individually-tailored prompts and suggestions to increase steps and movement and decrease SB time. We will iteratively design and test the intervention system with adult participants with the goal of having an interactive system that easily integrates into an individual’s daily living.