This week’s list of data news highlights covers July 31, 2021 – August 6, 2021 and includes articles on electronic identification programs and assessing the condition of infrastructure in major metropolitan cities.
1. Forecasting Technological Progress
Researchers at MIT have created a machine learning model to forecast the rate of progress of different technologies. The team classified 97 percent of inventions in the U.S. patent system into one of 1,757 groups based on the technological function they perform and scientific principles they use. The model then calculated how the impact of different patent groups changed over time to predict the annual performance improvement of different technologies.
2. Composing Music Using Cereal Boxes
General Mills, a U.S.-based food manufacturer, has launched an augmented reality app for consumers to play music from cereal boxes. Budding musicians can place pieces of Reese’s Puffs cereal into designated placeholders on the cereal box. The app then detects the placement of each puff to create a virtual music box.
3. Predicting COVID-19 Complications
Researchers at Mayo Clinic and nference, a U.S.-based biomedical data company, have created an AI system to predict the risk of COVID-19 complications. The team trained the system on electronic health records and clinical notes from 1,803 patients who were hospitalized for COVID-19. The system then identified connections between 20 risk factors reported by the Centers for Disease Control and Prevention (CDC) and 18 COVID-19 complications. Researchers found that hypertension is the most significant risk factor for developing complications from COVID-19.
4. Proving People’s Identities
Florida lawmakers are piloting a digital driver’s license and ID program. Under the program, Florida residents can prove their identity to law enforcement and participating retailers with an app that shows a scannable QR code or barcode containing identification information.
5. Tracking COVID-19 Vaccinations In Nursing Homes
The Illinois Department of Public Health (IDPH) has launched a data visualization tool that shows the weekly confirmed COVID-19 cases, deaths, and vaccination rates for residents and workers in 300 long-term care facilities across the state. Illinois law does not require long-term facilities to share vaccination data with IDPH so the tool uses public data that nursing homes report to the CDC.
6. Evaluating Genomic Tests for Prostate Cancer
Researchers at the University of Michigan have developed a test to determine the success rate of commercially available genomic tests for prostate cancer. The team used predictive analytics to determine the best treatment plan for patients with prostate cancer and compared resultant patient outcomes to similar plans and results from commercially available tests. The researchers found that one test, the Decipher Biopsy test, most closely matched their own predictions for the best treatment plan.
7. Predicting Suspicious Activity
The U.S. Department of Defense has developed an AI system that can detect suspicious changes in environments of interest around the world. The system analyzes satellite imagery and data from military-grade and commercial sensors to alert officials of suspicious activities. Officials can then analyze the images to determine the level of danger and a proper course of action.
8. Increasing Transparency in Police Responses
The Vermont State Police Agency has launched an open data portal to provide information on the use of force in police responses. The portal provides data from 2018 to 2020 on the type of force the officer used, the suspected contributing factor to the incident the officer is responding to, such as mental health or inebriation, and the time of each incident.
9. Predicting the Risk of Schizophrenia
Researchers at Baylor College of Medicine have developed a machine learning model that can predict the risk of schizophrenia from a blood sample. The team trained the model to identify variations in biomarkers between patients diagnosed with schizophrenia and people without the disease. In tests, the model identified patients with schizophrenia with 80 percent accuracy.
10. Assessing Roadway Conditions
The U.S. Chamber of Commerce Foundation has partnered with Roadbotics, a U.S.-based data visualization start-up, to assess roadway conditions in 20 major metropolitan areas. The team developed an AI system that identified variations, such as potholes and cracks, on 75 miles of roadways in each city. The system then ranked each city according to the results and found that Philadelphia, Jacksonville, and New York City have the best roadways.
Image credit: Flickr user Giuseppe Milo