RII Track 2 FEC: Multi-Scale Integrative Approach to Digital Health: Collaborative Research and Education in Smart Health in West Virginia and Arkansas

点击翻译按钮获取中文摘要

One potential approach to improve overall health outcomes and reduce healthcare costs is through the use of artificial intelligence techniques that can exploit the enormous amount of information embedded in huge and diverse health-related datasets. To leverage this large amount and large variety of information the project will develop new methods to address core research questions in artificial intelligence. Specifically, this project will develop and disseminate computational methods to maintain privacy while analyzing large datasets, develop and disseminate measures and methods to increase the transparency of data analysis and thus increase trust in the analysis results, and develop and disseminate methods to measure and reduce bias in big data sets. While privacy, transparency and bias reduction are important aspects of artificial intelligence in general, addressing these topics is especially urgent for health-related data and applications. The long-term goal is to accelerate decision making for smart health applications, through the development and application of advanced artificial intelligence techniques that can take advantage of available massive heterogeneous health-related datasets in an unbiased way. Successful realization of this goal will have significant broader impacts by spurring economic activity through improved workforce development in key technology areas of data science, artificial intelligence, and smart health.This project proposes a collaboration involving five partner institutions in West Virginia and Arkansas, and seven target primarily undergraduate institutions across the two states. Innovation in the project stems from the proposed techniques addressing difficult research challenges in artificial intelligence and data analytics, such as privacy-preserving data analytics, novel explanation-centric artificial intelligence techniques, multiscale approaches to exploiting diverse and massive health datasets using heterogeneous information network embedding, and implementation of new multi-view patient profile algorithms. Further innovation comes from the proposed non-trivial adaptations of these techniques for rapid and accurate decision making in smart health, by using large-scale computational deep learning techniques. High school students will be involved in STEM-related activities, while undergraduate and graduate students will be trained on leading-edge artificial intelligence and big data techniques and how these can be adapted for smart health applications. Workshops and summer schools will be used to educate students and faculty on research topics being studied in the collaboration, and to provide practical hands-on training on popular artificial intelligence platforms.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.