A Decade in Data Science and AI: Reflections on Building a Practice

A lot can happen in 10 years, particularly at the intersection of technology and science. In fact, the landscape for data science and Artificial Intelligence (AI) looks vastly different from even 1 or 2 years ago, demonstrating the fast-paced nature of modernization. New technologies require researchers to look at their work with a fresh lens to increase their efficiency, to solve problems in new ways, or to solve problems that were previously considered unsolvable—it truly is a brave new world of science. Despite recognized challenges, there are significant opportunities to innovate and utilize our multidisciplinary expertise to explore bold and original ideas to serve our clients and society better.

Over the past decade, RTI International has proactively focused on the field of data science and AI. I have had the pleasure to lead and witness the growth in this area, which includes assembling a strong team of 37 seasoned professionals (and counting) who have deep, broad, and nuanced expertise and significant accomplishments. We have solid leaders and national reputations. That experience, strength, and reputation are exactly what is needed to meet this incredible moment in the world, as digital transformation, data modernization, and generative AI are literally transforming our society as well as our federal government. 

Our Center for Data Science and AI is respected within RTI and in our industry for our data science acumen and achievements, for our collaborative nature, for being both a Center of Excellence and a Community of Practice, but also for our culture—our team cohesion and our people-centered orientation. We have fun, and we help one another; we learn from one another; and we have each other’s backs. We cheer each other on. These are not small things. I have found that whatever we do in life, in the final analysis, it’s always about people, and I’ve tried to lead with that as my compass.

Although there may be good reason for simultaneous excitement and caution about incorporating rapid technological change into our work, I remain convinced that technology can be used responsibly and equitably to solve problems and improve the lives of all Americans.  

A Decade of Data Science and AI

Looking back on where we started, it is incredible to see how over the years we have leveraged the power of data science to solve complex scientific and societal problems and to improve public health and well-being. The following are just a few examples.

Partnership with the Department of Justice: Improving Arrest-Related Death Tracking

In 2015, we partnered with the Department of Justice (DOJ) to improve the system that tracked and counted arrest-related deaths.  The data were previously incomplete due to voluntary participation in the system by law enforcement agencies, and RTI was asked to redesign the process using online data in the form of news articles. By combining our expertise in data engineering, natural language processing, and machine learning, we created a process that dramatically reduced the number of online articles that needed manual review, while maintaining a high accuracy rate. Our process led to the most comprehensive official effort to record the accurate number of deaths at the hands of American law enforcement and provide the “national, consistent data” described by Loretta Lynch, the United States Attorney General at the time. Because of these and related efforts, we were then invited to serve on President Obama's Data-Driven Justice Initiative. 

Addressing the Vaping Crisis: Utilizing Natural Language Processing Algorithms

In response to the nation’s vaping crisis in 2018, our team used natural language processing algorithms—developed under a previous grant from the National Institutes of Health (NIH)—to show that the majority of the audience following JUUL’s marketing on social media were underage youth. This generated a lot of buzz in the scientific community and among national news outlets, which led to consulting opportunities with several State Attorney Generals offices and the deployment of a public health surveillance tool—the Electronic Nicotine Delivery System (ENDS) Tracker—to help state and local health departments monitor emerging vaping brands and to prepare proactively for “the next JUUL.”

Enhancing Healthcare Networks: RTI's Collaboration with CDC

From 2018 through 2021, RTI assisted the Centers for Disease Control and Prevention (CDC) in examining how hospital acquired infections spread through health care networks—including how to detect these pathogens and prevent their spread. With that experience, we were able to pivot quickly during the COVID-19 pandemic, answering the call from the North Carolina Department of Health and Human Services and the North Carolina Governor’s Office to adapt our simulation models to forecast hospital bed and intensive care unit capacity. The models provided valuable insights for state leadership to make weekly, data-driven public health policy decisions during a tumultuous period. These data translated to life-saving information for people across the state.

Streamlining Public Comment Review: RTI's Collaboration with CMS

In 2022, a collaboration with the Centers for Medicare & Medicaid Services yielded an improvement to the efficiency and timeliness of public comment review of new health policy rules. The RTI SmartReview data science tool used an automated data pipeline, natural language processing, and machine learning to select relevant comments and integrate organized comment data into dashboards quickly, accurately, and automatically. RTI SmartReview assessed more than 30,000 public comments and identified and organized relevant comments for consideration within 2 days of the comment period closing. This effort won the 2022 FedHealthIT Innovation Award, recognizing innovation in data science.

Investigating Long COVID: RTI's Collaboration with NIH and UNC Chapel Hill

From 2022 to the present, with additional support from NIH—through the RECOVER Initiative: Researching COVID to Enhance Recovery—our team has been using machine learning and applying a variety of advanced methodological techniques to understand how to predict, prevent, and treat Long COVID with electronic health record data. In collaboration with scientists from the University of North Carolina at Chapel Hill, RTI data scientists are contributing analyses for a variety of Long COVID questions, including investigating the relationships between vaccination and Long COVID, comorbidity indices, and reinfection. Analyses have been included in reports to the National Heart, Lung, and Blood Institute and to the broader NIH-RECOVER community; quoted in the media; and presented at the White House.

Harnessing Generative AI to Advance Research at RTI

Finally, in 2023, with the increased use of generative AI and large language models—particularly with the introduction of tools like ChatGPT—our data scientists embarked on a journey to see how these methods can be applied in research. These tech-forward approaches have positioned us to support federally funded projects better and help our federal agency partners understand how to apply AI effectively, equitably, and responsibly.     

A Season of Celebrating Our Data Science and AI Initiatives

We’ve been hosting several events to celebrate a decade of success—from creating a team from scratch to building a business and brand in a new and exciting practice within a large contract research organization dedicated to improving the human condition. To commemorate the occasion, we were fortunate to have been joined by a special guest speaker at the April 23 edition of our ongoing internal Data Science Lunch and Learn series, highlighting their own journey through data modernization and digital transformation. It was an occasion to look back and reflect, and to learn from the experience of our governmental partners, as we forge headlong into the next decade of data science and AI.

When we officially started in 2014, we were still learning what data science was and what it could and should be at RTI and in public sector research. In fact, it was still new everywhere at that time, and it seemed like everyone was attempting to figure it out. Considering that many, if not most, new things in the world don’t succeed, there was no reason to think our data science and AI initiatives would be any different. But with grit, determination, and a “never give up” attitude, we took it one day and challenge at a time and learned as we went. Before long, we were regularly solving problems and creating impact, and there was no looking back. I am proud thinking about how we navigated the inevitable high-highs and low-lows of any new venture—through it all, I took heart in remembering that “nothing in life that’s worth anything is easy.” 

Shaping the Future of Data Science and AI

From the social sciences and public health to the biological sciences, our team continues to shape the future of data science, realizing its mission for solving important national problems, improving our local communities, and transforming research.  We are actively working on defining policy and implementation tactics for Responsible AI, contributing to applied research in machine learning and natural language processing, bringing new sources of real-world data to research and policy, discovering how generative AI and large language models can be augmented with existing forms of automation and machine learning in our project space, and building and deploying AI and other modern automation systems into our projects and deliverables.

This decade, we navigated some of the world’s most difficult challenges—at least in our lifetime. During the pandemic, we witnessed how scientists and policymakers had to respond quickly to address a public health emergency. It was during that time that data-driven decision making got to show its worth. From the federal government’s commitment to modernizing to make better use of its data to our own work saving lives close to home, data science showed its ability to solve a human problem. Our team remains devoted to innovation, to problem-solving, to collaboration and mentoring, and to navigating uncertainty—as it will be ever present. With our eyes on the upcoming 10 years, we will continue to solve problems once considered unsolvable, shaping the future of data science in the research community, the federal government, and beyond. 

Learn more about how RTI continues to adapt to a changing data science and AI landscape.