In This Story
Data fuels AI. Sound design of experiments, and, more broadly, data collection methods for controlled experiments, clinical trials, and crowdsourcing, whether extracting data from electronic health records (EHRs), government sites, national databases, surveys, other internet sources, or newly collected online, is essential for obtaining unbiased data for responsible decision-making and AI.
George Mason University's Department of Statistics stands at the forefront of innovation, including in the realm of design of experiments (DoE) — a critical component of research and statistical application. This field is pivotal in ensuring that data collection methods are optimized to solve specific problems, making unbiased comparisons, and detecting treatment differences more effectively. The department not only includes a significant number of experts specializing in DoE but also experts on extracting data from real-world evidence (RWE), such as EHRs, study protocols, and crowdsourcing.
The DoE experts include William Rosenberger, the distinguished university professor with international honors. He is known for his work on clinical trials, having authored two Wiley books on randomization in clinical trials, was recently named the 41st Fisher Memorial Lecturer, and served on the international advisory board for the European Union and also DSMBs for clinical trials sponsored by both government and industry.
John Stufken, the professor and a renowned world expert on optimal industrial experiments, also has two books on DoE, published by Springer and Chapman & Hall / CRC. His recent work integrates subsampling into DoE. Nicholas Rios, a dynamic young tenure-track assistant professor, also specializes on DoE. His work includes a notable NSF grant project where he designed an experiment comparing faculty training methods, including virtual reality, to traditional approaches. This innovative application underscores the department's role in pioneering new methodologies and fostering interdisciplinary collaboration. Experimental design is fundamentally about collecting data in the most appropriate way to address specific research questions.
Rios explained, “Experimental design involves creating unbiased comparisons and detecting differences between treatments, such as medications, with high accuracy. Key considerations include determining the number of data points, selecting appropriate samples, and addressing confounding effects like age in treatment studies. These principles are crucial in fields ranging from clinical trials to A/B testing and subsampling, where the goal is to derive meaningful insights from complex data. Of course, the department's faculty are instrumental in its success." Rios emphasizes the department's engagement with emerging problems in data analysis, particularly in the context of AI and large data volumes. His work on NASA's remote sensing data analysis showcases the department's innovative approach to subsampling, enabling complex modeling on standard laptops rather than requiring high-performance computing resources.
Along with Rios, John Stufken, and William Rosenberger, other faculty, including Anand Vidyashankar, Lily Wang, and Jiayang Sun, have been contributing their extensive statistical expertise to important, large collaborative projects, where the data collection and handling from RWE can be messy. They include those related to security and privacy (Vidyashankar), statistical imaging (Wang), interpretable models for Heart transplants, EHRs, and crowdsourcing (Sun). All have maintained an excellent funding history with federal agencies and/or industries.
Looking ahead, the Department of Statistics is poised to explore new frontiers in experimental design, particularly in AI and data volume analysis. “I encourage aspiring statisticians to delve into these emerging areas and contribute to an ongoing commitment to innovation and collaboration,” said Jiayang Sun, department chair. “By embracing new methodologies, our researchers aim to continue to lead the field. Please check out our faculty profiles, including young faculty and newly arrived faculty. We will be hiring again this season.”