This week’s list of data news highlights covers June 26, 2021 – July 2, 2021 and includes articles on predicting immunotherapy responses and analyzing dark matter in between galaxies.
1. Democratizing Municipal Data
The City of Los Angeles has launched a new website, called “Know Your Community,” for residents to access municipal data. The site uses data available on the city’s open data portal and includes maps on 311 service requests, healthcare information, and public safety. Residents can use the portal to download data and view content for a specific neighborhood or demographic group.
Researchers at Pennsylvania State University have developed a machine learning model that can map dark matter between galaxies. The team trained the model on data from galaxy simulations built to resemble the Milky Way, and tested it on data from 17,000 galaxies within 3.26 light-years of the Milky Way. The model accurately mapped dark matter filaments that connect the Milky Way to a nearby galaxy named Andromeda.
3. Projecting the Impact of Sea Level Rises
Researchers at the National University of Singapore and Deltares, a research institute in the Netherlands, have partnered to map land vulnerable to sea level rises. The team used lidar data from a NASA satellite to find land less than 2 meters above sea level. They then combined the results with population data to find that 410 million people live in affected zones.
Music producers in South Korea have created an AI system that enables K-pop musicians to perform using a virtual avatar. Producers used the system to create an 11-member girl group called Eternity, where musicians can each use as many as 10 artificial characters to perform and conduct video interviews.
5. Identifying Drought-Resistant Crops
Researchers at Cornell University have developed a nanoscale sensor that can assess crop health by measuring a plant’s water availability. Researchers inject sensors into a plant, and the sensors expand or contract depending on water availability. Researchers can then pick up this movement with a spectrometer instrument and convert the data into water potential measurements. Farmers can use the technology to identify and develop crops with drought-resistant genes.
6. Mapping Cool Walking Routes on Hot Days
Barcelona city officials have released a new app that maps the coolest walking route between two points on hot days. Geographic analysts first created a 3D model of the city with lidar data, and then manually added municipal data on drinking fountains and shelters. City residents can use this app to avoid extreme heat during increasingly common heatwaves.
7. Analyzing Sports Fans’ Emotions
Researchers with Realeyes, an international ad testing company based in the United Kingdom, have developed an AI system that can detect sports fans’ moods when watching games. The system analyzes facial expressions to detect emotions based on movement in the eyes, eyebrows, nose, and mouth. For example, the team found that fans express the most shock during Formula 1 after detecting differences in viewers’ eyes and jaws.
Researchers at U.S. technology company Impact Observatory and U.S. geographic information system software company Esri have created an AI system that mapped the planet’s land cover in less than a week. The team trained the system on data from five billion image pixels of the planet. The system then used 400,000 satellite images of Earth to produce a global land cover map. Researchers can use the system to quickly update land cover maps before changing environments make the satellite data outdated.
9. Predicting Immunotherapy Responses
Researchers at the Eindhoven University of Technology in the Netherlands have developed a machine learning system that can predict a patient’s responses to immunotherapy. The team trained the system on datasets from historical patient data, typical immune responses, and RNA-sequencing, a method of identifying biomarkers that signify the constitution of a tumor’s microenvironment. The machine learning system predicted patient responses to immunotherapy treatments better than previous prognostic assessments used in clinical settings.
10. Forecasting Fire Flashovers
Researchers at the National Institute of Standards and Technology and Hong Kong Polytechnic University have created a machine learning model that can predict flashovers, a phenomenon where all combustible material in a fire ignites simultaneously. The researchers trained the model on more than 5,000 simulated fires in a virtual three-bedroom house. They then tested the model’s ability to predict flashovers in 13 real housefires in laboratory experiments. The team found that their model could predict a flashover in large rooms 30 seconds before ignition.
Image credit: The National Institutes of Health