This week’s list of data news highlights covers April 18-24, 2020, and includes articles about automating the detection of plastic in oceans to predicting when allergies will flare-up.
Researchers from Boston Children’s Hospital have developed Covid Near You, a crowdsourced tracker of individuals experiencing symptoms of the disease. Over 800,000 people have provided their symptoms and zip code to the tracker, allowing the researchers to map the prevalence of symptoms by location. The maps show recent upticks in Denver, California’s Central Valley, and Dayton, Ohio.
Researchers from Plymouth Marine Laboratory in the United Kingdom have developed an AI system that can analyze satellite imagery to detect sea plastic. The researchers trained the system using images taken by European Space Agency satellites. The system correctly classifies items as plastic, as opposed to elements such as seafoam, with 86 percent accuracy.
CareATC, which provides on-site healthcare services to organizations, is using an AI system to identify their client’s employees that have a high risk of severe illness if they contract COVID-19. The system uses data on an individual’s health, emergency room use, and zip code. CareATC contacts flagged individuals to provide them information on proper hygiene and what to do if they experienced flu-like symptoms.
Researchers from IBM have developed a new tool that can predict when an individual will display allergy symptoms with up to fifty percent more accuracy than systems that solely use pollen to make predictions. The researchers trained the tool’s model on data from patients, their location, and weather attributes such as temperature, wind, and dew point. The combination of geographical and weather data helps the model predict when flora will grow and produce allergens in an area.
Researchers from the U.K.’s National Health Service (NHS) and the University of Cambridge have developed an AI system that can predict the demand for intensive care units and ventilators to treat COVID-19 at individual hospitals. The researchers trained the system, which NHS is trialing at four hospitals in England, on the deidentified data of 4,000 patients. The system allows hospitals to run simulations of different scenarios, such as the effects of a change in the demographics of admitted patients.
Researchers from the University of California have developed an AI system that can detect real Twitter users from bot accounts. The researchers trained the system on nearly 12 million tweets from 8,500 human and bot accounts and the system found that humans replied to other tweets roughly five times more often than bot accounts. In addition, the system, which had a 97 percent probability of correctly classifying an account, found that bots often tweeted in distinct time intervals.
Intel and Accenture are using an AI system to monitor the health of coral reefs in the Philippines. The system uses an artificial reef, equipped with cameras, to detect, classify, and photograph fish as they swim by. The system sends the data to researchers in real-time and has taken more than 40,000 images.
The European Commission has launched the COVID-19 Data Portal to increase access to datasets for research combating coronavirus. Researchers can both upload and access data from the portal, which will store data such as DNA sequences, protein structures, and data from clinical trials.
Innowatts, a startup based in Houston, has developed a system that uses machine learning to better forecast utility companies’ energy demand. The system uses machine learning algorithms to analyze data from 34 million smart energy meters to make long and short-term forecasts. The forecasts help utility companies reduce the possibility of brownouts or blackouts.
Researchers from the Battelle Memorial Institute, a nonprofit science and technology development firm, developed a brain-computer interface that restored movement and the sense of touch to a man with a severe spinal cord injury. The system analyzed brain activity data from an implant and decoded the signals into muscle movements, allowing the individual to perform 20 different types of hand grips and even play the video game Guitar Hero. The system has also helped the individual overcome the inability to feel objects in his hand through an armband on his bicep that vibrates when his hand receives sensory input.