This week’s list of data news highlights covers July 10, 2021 – July 16, 2021 and includes articles on fighting wildfires and designing wearable technology for the U.S. Olympic team.
1. Spotting Anemia in Pictures
Researchers at Brown University have created an AI system that can spot anemia from pictures of the inside of a patients’ lower eyelid. The team trained the system on pictures of eyelids and medical data from 142 patients. During tests, the system diagnosed anemia with 72.6 percent accuracy.
2. Incorporating Wearable Technology into Olympian Uniforms
Ralph Lauren, a U.S.-based fashion brand, has unveiled a denim jacket that uses sensors to monitor and optimize its wearer’s temperature. If the sensors detect overheating, a battery-powered fan located on the back of the jacket will blow cold air onto the wearer. Flag bearers for the U.S. Olympic team will wear the jacket during the Olympic and Paralympic Opening Ceremonies.
Researchers at Stanford University have developed a machine learning model that can predict a patient’s biological age by identifying markers of inflammation associated with heart disease. A patient’s biological age can differ from their chronological age due to inflammation and other health considerations. The team trained the model to identify protein markers that signify the presence of inflammation on clinical data and blood samples from 1,001 patients.
4. Stargazing with Augmented Reality
Au Diable Vert, a Canadian outdoor recreation center, has launched the world’s first augmented reality planetarium experience in Sutton, Quebec. At the experience, known as Observétoiles, visitors point their smartphone at the night sky. An app then identifies constellations, tours the solar system, and superimposes original 17th-century illustrations against their modern counterparts in the sky.
U.S.-based start-up Wonder Inventions has created an AI system that can take a composed piece of music and reconstruct it so it fits with the length and pace of a video. The team broke down original songs and out-of-copyright classics into smaller musical units called morphones and trained the system to reconstruct musical pieces that were similar to the original while matching the speed and tone of a short video.
6. Fighting Wildfires with Data
Researchers at the University of California, San Diego have built an interactive fire map to inform firefighting efforts. The map uses data from satellites, wind patterns, utility service operations, weather stations’ live streams, and past blazes to create a real-time infographic of wildfire risks across the state. Fire officials can use the map to direct resources and evacuate local communities during wildfires.
7. Predicting the Severity of COVID-19
Researchers associated with the National Institutes of Health have built a machine learning model that can predict a patient’s chance of developing severe COVID-19 symptoms from data collected during the first 24 hours of their hospital stay. The team trained the model on clinical data from 174,568 patients who tested positive for COVID-19. The model then identified six factors that increase a patient’s chance of developing severe symptoms: age, sex, race, liver health, weight, and dementia.
8. Monitoring Bridge Integrity
Researchers from Imperial College London have installed a 3D-printed bridge over the Oudezijds Achterburgwal canal in Amsterdam. The bridge contains sensors that transmit real-time data on the bridge’s integrity and pedestrian traffic to the team. Researchers will use the data to build a digital twin of the bridge and monitor its health in the long term.
9. Ensuring Safe Human and Robot Interactions
Researchers at MIT have developed a machine learning model that can teach robots how to safely help humans with everyday tasks like getting dressed. The team built upon existing models by training the robot on human reactions. In tests, the team’s robot successfully dressed a human into a vest.
10. Improving the Flavor of Plant-Based Hamburgers
Swiss fragrance company Firmenich has developed an AI system that can generate ideal flavor combinations for plant-based hamburgers. The system uses data on consumer preferences, technical or regulatory constraints, and potential ingredients to suggest flavors for plant-based meat. The company recently launched the first flavor the system created, a lightly grilled beef flavor that can improve the taste of plant-based foods.
Image credit: Flickr user alistair.pott