This week’s list of top data news highlights covers June 7th 2025 to June 14th, 2025 and includes articles on tracking professional golf games with AI and exploring the mining history of the lost town of Treadwell, Alaska using augmented reality.
Naver, a South Korean Internet company, has developed a digital twin platform to support urban planning in the holy cities of Mecca, Medina, and Jeddah in Saudi Arabia. The platform uses ultra-high-resolution 3D imagery and real-time data streams to digitally map approximately 6,800 square kilometers and more than 920,000 buildings, enabling terrain analysis, sunlight exposure modeling, disaster simulations, and construction compliance checks. Saudi planners are now using the platform to evaluate zoning, infrastructure, and disaster preparedness with spatial precision.
2. Improving Hurricane Forecasts
DeepMind, an AI company based in London, has partnered with the U.S. National Hurricane Center to improve predictions of hurricane paths and intensity. The team trained their specialized model on data from nearly 5,000 storms over the past 45 years, combining observational records from satellites, balloons, ships, and radar. The system produces fifteen-day forecasts that match or exceed traditional models in tracking storm direction and strength, and is being tested alongside human forecasters in the current Atlantic hurricane season.
3. Tracking Professional Golf Games
The U.S. Golf Association, based in the United States, has deployed an AI-powered ball-tracking system, called ShotCast, to monitor every stroke during the Men’s U.S. Open. ShotCast, developed with Cisco, combines ball-tracking sensors and generative AI to analyze data on approximately 30,000 shots, including speed, spin, and trajectory. The system has been used to deliver shot-by-shot updates about more than 150 players during the tournament.
4. Reconstructing Black Hole Data
The Morgridge Institute for Research, based in Wisconsin, has developed an AI system that helps astronomers recover better images of black holes from blurry telescope data. Normally, computers discard much of the raw data collected by the Event Horizon Telescope because it’s hard to interpret. To solve this, researchers trained their AI system using millions of simulated black hole images. This training helps the AI recognize what real black holes should look like, allowing it to extract meaningful patterns from noisy data. Using this method, scientists were able to generate a new image of the black hole Sagittarius A*, revealing that it may be spinning near its maximum speed and pointing its rotation axis toward Earth.
5. Automating Multi-Step Workflows
Mechanize, based in San Francisco, has built AI agents capable of completing complicated workflow tasks from start to finish. The agents are trained in virtual office environments—complete with browsers, email inboxes, coding tools, and messaging platforms—where reinforcement learning is used to reward successful task completion and penalize failures. In early tests, the company’s AI agents could successfully replicate the workflows of junior back-end programmers, completing tasks such as writing and debugging code and coordinating over Slack in a simulated development environment.
6. Testing Insulin Adjustments
Researchers at the University of Virginia have developed digital twins of people with type 1 diabetes to help them optimize insulin dosing. By creating replicas of 72 patients using continuous glucose monitors and automated insulin delivery systems, participants can test how different activities and insulin strategies would affect their glucose levels, allowing them to simulate insulin adjustments before making real-world changes. Users increased their time in a healthy blood sugar range from 72 percent to 77 percent.
The Office of the Chief Technology Officer, DC Health, and the Department of Energy and the Environment in Washington, D.C., have developed a platform consisting of D.C. city maps to visualize how neighborhood conditions relate to health access. It analyzes over 40 indicators—including proximity to schools, public transit, grocery stores, housing, and medical care—using city agency records compiled through Open Data DC. The map’s have already revealed sharp spatial differences in access to food, transportation, housing quality, and healthcare across D.C. neighborhoods.
Chronicle, a California-based startup, has developed an AI-powered tool that helps users turn documents, PDFs, or web links into scrollable, web-style presentations. Instead of using traditional slides, the platform breaks content into blocks for text, charts, images, and videos. The system uses AI to analyze the input, suggest a narrative structure, and apply layout and design rules so that users don’t need to format slides manually.
9. Rebuilding a Mining Town with AR
The Treadwell Tour Company, based in Alaska, has launched an augmented reality platform that reconstructs the lost mining town of Treadwell for visitors walking through its ruins. The system overlays 3D animations, historical photographs, and virtual recreations of machinery and buildings onto real-world sites, using licensed archival images and precise geolocation. Visitors can view digital versions of long-destroyed structures, explore animated storylines, and walk through a historically accurate landscape without disturbing the surrounding forest.
10. Improving Cardiac Surgery Recovery
The Icahn School of Medicine at Mount Sinai, based in New York, has developed an algorithm to recommend personalized insulin doses for patients recovering from cardiac surgery. Trained on over 6,000 ICU cases, the system outperforms clinicians in accuracy and safety. This has enabled more precise insulin management during the unstable postoperative period, reducing risks linked to hyperglycemia and improving early recovery outcomes.
