From Hidden History to Robot Surgery

Newswise — Today’s scientists have the ability to produce and access more information than at any other point in history, from atlases of gene expression to global climate data to digitized historical archives. This sea of information can be used to answer previously unsolvable questions—if researchers can find ways to harness it. This was the challenge that brought together hundreds of scientists this January to attend Utah Data Science Day 2024.

Although the attendees shared a passion for data, they came from a remarkable variety of scientific backgrounds and career stages. This was no coincidence, according to Penny Atkins, PhD, associate director of the One Utah Data Science Hub. “Data science is not something that can happen without collaboration,” she said, adding that the field uniquely requires cross-talk between data analysis experts and scientists in a variety of applied research areas. The attendees’ diversity of thought was evident from the start, as the three research talks that kicked off the event demonstrated ways to use big data to solve three very different problems.

Using VR to help surgical robots learn

Alan Kuntz, PhD, assistant professor in the Kahlert School of Computing and the Robotics Center at the University of Utah, asked how we might relieve the burden on overworked surgeons by training surgical robots to assist them more effectively. A major hurdle is that, unlike most learning robots, surgical robots can’t be trained using real-life data—there are no “practice surgeries” on real people. His proposed solution is to use a physics simulator combined with demonstrations from surgeons to teach the robots to do comparatively simple tasks, like hold a layer of tissue out of the way of a human surgeon. “This ends up working really well,” Kuntz said. Using simulated data, the team has already trained a robot to autonomously perform multiple assistive surgical tasks in the real world, outside of simulations.

Tracking an invisible epidemic

The next talk zoomed out from the operating room to what Katharine Walter, PhD, assistant professor of epidemiology at U of U Health, calls an “invisible epidemic”: Valley fever, a fungal disease that spreads through soil and dust and can cause pneumonia when people inhale spores. Valley fever often goes undiagnosed, and nobody knows how prevalent the fungus is in the soil in Utah. Walter’s team is collecting environmental samples and sequencing fungus genomes to understand where the fungus is now. By combining her results with large-scale climate and biodiversity datasets, she aims to predict where it might spread as the climate changes. “There are many opportunities to collect new types of data that can directly inform public health,” Walter said of her work.

Finding history’s hidden influencers

Peter Roady, PhD, assistant professor of history at the U, and John Gordon, program manager for the SCI-HUM Collaboration Initiative, presented an online platform to help researchers find relationships in historical records. “It has never been easier for humanities researchers to amass vast archives, but as we know, gathering data is only the first step in the research process,” Roady said. “That data has to be analyzed, and that analysis has become increasingly overwhelming.” Gordon and Roady’s platform extracts names from uploaded documents and maps connections between individuals, identifying key players in historical movements who had previously been overlooked. With this tool, “We can enable researchers to see into a far larger archive than they could ever achieve on their own,” Roady said.

Artificial intelligence for better data science

The conference’s keynote talk dived deeper into AI as Susan Gregurick, PhD, associate director for data science and director of the office of data science strategy at the National Institutes of Health, gave an overview of the NIH’s strategies to support better data management, make clinical data more accessible, and promote collaboration. Artificial intelligence was a throughline in many of the initiatives she outlined, from research to make real-world diabetes data understandable to intelligent data processing tools, to a consortium to use AI and machine learning to advance health equity and diversity.

While the problems the researchers hoped to address with their research were often serious, weighty issues, the overall atmosphere was one of optimism and excitement. Participants ranging from undergraduate computer scientists hunting for summer internships to study coordinators in oncology looking for collaborators specializing in metabolism set the large hall abuzz with conversation as they made new connections. Warren Pettine, MD, an assistant professor at Huntsman Mental Health Institute, summed up the mood at his poster on simulating how the brain makes decisions: “I like data, and I like science. So why not do data science?”