This week’s list of data news highlights covers May 22, 2021 – May 28, 2021 and includes articles about supercomputing research and machine learning in the healthcare industry.
The Utah Department of Transportation (DOT) has partnered with Blyncsy, a startup in the state, to pilot a tool that uses mobility data to determine roadway conditions. The tool, called Payver, collects and analyzes crowdsourced images and videos from dash cams in cars to offer insights to government. For its pilot program, Utah DOT will use Payver in Salt Lake City to monitor roadway paint lines, which need to be visible for autonomous cars to navigate effectively.
Stafford County in Virginia and the Center for Innovative Technology, a nonprofit focused on accelerating innovation in Virginia, have launched a testbed in the county that has a connected network of sensors to test emerging technologies, such as drones and smart flood control. They hope to improve policymakers’ ability to make decisions, such as when best to implement a shelter, or evacuate citizens, or close roads that are prone to flooding.
Google and U.S.-based hospital chain HCA Healthcare have teamed up to develop algorithms that can improve medical care. The two companies expect better use of data will allow them to increase hospital operating efficiency, monitor patients, and provide medical guidance for doctors. HCA Healthcare previously used an algorithm for patient care during the COVID-19 pandemic and found that survival rates for critically ill patients increased after implementation.
U.S. AI startup Merlin Labs has partnered with aviation company Dynamic Aviation to outfit and test autonomous pilot systems on aircrafts of various sizes. Many companies are developing systems to autonomously fly planes that have a fixed wing length, but Merlin Labs is developing its system for use on planes of any wing size. Most recently, the startup tested its technology on 55 large King Air planes. The team envisions the first use of its new technology will be for cargo, but that use could expand to military transport and firefighting flights.
Researchers at the University of Illinois Urbana-Champaign and California Institute of Technology have used a supercomputer to better understand why some tsunamis triggered by earthquakes are so devastating while others are not. The team modeled tsunamis stemming from earthquakes using the Blue Waters supercomputer at the National Center for Supercomputing Applications and found that earthquakes with particularly fast movements can trigger significant tsunamis because of the way they shake coastal land, displace water, and move waves. The researchers hope geologists will use the results to revisit tsunami risks and employ more targeted controls to mitigate damage in narrow bay regions.
Rover Development, an engineering startup in Minneapolis, has developed a battery-free sensor for bicycle tires known as the PSIcle sensor. Users can ascertain a tire’s measurements by holding their smartphone next to the sensor. This will open the PSIcle app and display measurements for the pressure and temperature of bicycle tires.
Geico Insurance, which is the second largest insurance company in the United States, has partnered with Tractable, a damage appraisal company that uses AI to speed up recovery, to expedite car repair estimates. Under this partnership, Tractable algorithms will analyze pictures of cars to determine the extent of damage and then relay a repair estimate to the auto-body shop within minutes. A human can then review the damages and provide an estimate more quickly.
Flawless AI, an AI startup in London, is using technology to improve how movie producers dub foreign language films. The tool works by analyzing how actors move their mouths in movies and changes their lip movements in the video to better match pre-recorded audio. For example, an actor speaking English in an American movie could be made to appear as if they are speaking Japanese.
The city of Chicago has partnered with the University of Chicago Crime Lab to launch a data dashboard that provides public access to real-time violence data. The tool uses data on violent crimes such as shootings, aggravated assaults, domestic violence, and human trafficking incidents from the Crime Lab as well as from city departments and victim services. So far, the data has illustrated the disparate impact of violence in Chicago both geographically and demographically.
Researchers from Duke University have developed an AI system that can analyze stool samples for signs of gastrointestinal issues. The researchers designed the technology to be installed in the pipes of existing toilets, where it takes pictures of stools that pass through and analyzes them as loose, normal, constipated, or bloody using an algorithm trained on 3,000 images. In tests, the tool accurately classified 85 percent of stools and correctly detected blood 76 percent of the time.
Image credits: Flickr user Andrew E. Cohen