This week’s list of data news highlights covers April 4-10, 2020, and includes articles about tracking whales using satellites and simulating how coronvirus could spread in grocery stores.
Researchers from Google have developed an AI system that can generate syllables to fill in missing audio during video calls. The audio in online calls is transmitted over the Internet in packets, but the packets can arrive late or not at all, causing glitches in calls. Google trained a neural network on a dataset of 100 humans speaking 48 different languages, allowing it to generate the most likely syllables when a packet is lost.
Facebook is using anonymized data from users who opt-in to location tracking to create tools to help public researchers track social distancing. For example, the tools include maps that show if individuals are generally staying close to home and co-location maps that indicate the likelihood that individuals in disease outbreak areas will come in contact with other populations.
Researchers from DeepMind have created an AI system that can predict the movement of glass molecules during their transition between liquid and solid states. The researchers trained the system by representing molecules and their relationships as nodes and edges on a graph, allowing the system to predict molecule movement. The system’s predictions achieved a correlation of 96 percent with ground truth data over short time scales.
4. Using Mobile Data to Enable Contact Tracing
A group of more than 100 European researchers have developed code for an app that can enable contact tracing. The system uses Bluetooth signals in mobile phones to detect if individuals are close enough to each other to spread COVID-19. If an individual is later confirmed to have the virus, it sends alerts to other users who could have contracted COVID-19 due to close proximity.
5. Tracking Whales from Space
Researchers from Charles Stark Draper Laboratory and the New England Aquarium are using data from satellites, sonar, radar, human sightings, and ocean currents to train a machine learning algorithm to predict the locations of whales. The algorithm could help authorities make decisions to protect the whales, such as altering shipping lanes and speeds.
6. Crowdsourcing COVID-19 Data
Researchers from Cambridge University have started a crowdsourcing project to create an AI system that can analyze audio to detect if an individual has COVID-19. Members of the public can submit audio of themselves coughing, breathing, and speaking to the project’s website, which also collects information on an individual’s age, gender, approximate location, and if they have had COVID-19. Over 1,200 people provided recordings on the project’s first day, including 22 with COVID-19.
7. Collecting Data to Track the Spread of COVID-19
Researchers from Carnegie Mellon University have partnered with Google and Facebook to collect data to predict the spread of COVID-19. The firms are asking volunteers who use Facebook or Google’s Opinion Rewards app, which lets individuals answer surveys for app store credit, if they are experiencing coronavirus symptoms. The researchers from Carnegie Mellon University hope the data will allow them to create county-by-county projections concerning COVID-19.
The Jacksonville Transportation Authority is using autonomous vehicles (AVs) to transport COVID-19 tests from a drive-thru testing site to a Mayo Clinic processing laboratory. The vehicles, which are from Beep and NAVYA, transport up to 150 tests a day. The vehicles have allowed the Mayo Clinic to reassign personnel to other tasks.
9. Determining the Origins of New York City COVID-19 Cases
Researchers from New York University and the Icahn School of Medicine at Mount Sinai in New York City have separately analyzed coronavirus genomes, finding that the virus entered the New York area mostly from Europe. The researchers from New York University also found that some of the viruses had unique mutations, indicating that the virus had likely been spreading from asymptomatic individuals in the New York area.
Researchers from multiple Finnish organizations have used a supercomputer to model how coronavirus spreads in grocery stores. The researchers modeled how airborne particles from an individual’s cough could spread and linger in an aisle, finding that other individuals could come into contact with the particles minutes after the original cough. The researchers conducted the simulation using a supercomputer from CSC, a Finnish state-owned science and technology company.