CRI: CI-EN: Collaborative Research: mResearch: A platform for Reproducible and Extensible Mobile Sensor Big Data Research

点击翻译按钮获取中文摘要

The Center of Excellence for Mobile Sensor Data-to-Knowledge (MD2K) has developed open-source software for smart phones and cloud. Scientists use MD2K software to develop and test algorithms to monitor health, wellness, and work productivity via wearable sensors. The mResearch project is aimed at assisting Computer and Information Science and Engineering (CISE) researchers. The mResearch project will significantly enhance MD2K software and integrate Internet-of-Things (IoT) devices. The enhanced MD2K software will accelerate research in sensors design, mobile computing, privacy, analytics (especially machine learning and deep learning), and visualization. mResearch will enable CISE researchers to easily deploy their contributed software in scientific studies for health, smart homes, and workplace. The resulting discoveries and tools will help individuals improve their health, wellness, and work productivity.MD2K has developed open-source mobile sensor big data software platforms mCerebrum for smartphones and Cerebral Cortex for the cloud. This scalable and generalizable infrastructure is used for collecting, analyzing, and sharing high-frequency, mobile sensor data and associated labels in the context of scientific field studies. In particular, it supports the development and validation of models and algorithms for inferring markers of health, wellness, and productivity, and their associated risk factors. It has already been used at eleven sites across the country to collect over 300 terabytes of mobile sensor data in the field setting from over 2,000 participants. It has resulted in new computational models for the detection of conversation, smoking, eating, craving, stress, and cocaine use. The mResearch project is making five significant infrastructure enhancements to the MD2K infrastructure to assist CISE researchers in mobile sensor development, mobile computing, privacy, analytics, visualization, and participant engagement. First, it will enable data analytic workflow management across multiple layers of the system to enable reproducible and extensible experimentation. Second, it will allow encapsulation of data sources to provide convenient and responsible access to them in data analytic workflows. Third, it will facilitate cloud-assisted complex, real-time analytics for personalizing mobile interventions and improving engagement. Fourth, simulators will be developed with the ability to feed stored data into the platform at various points to enable research on system components and properties such as data compression, transfer and storage, as well as the scalability of data analytics. Finally, Internet-of-Things (IoT) devices and services will be integrated. With these five enhancements, the MD2K software will provide a complete, open, and modularized architecture. It will include all aspects of sensor data collection, data processing algorithms, cloud-based machine learning, and IoT integration. The enhanced MD2K software will facilitate reproducible and extensible CISE research with high-frequency mobile sensor data.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.