Home BlogWeekly News 10 Bits: the Data News Hotlist

10 Bits: the Data News Hotlist

by Joshua New
Kuala Lampur

This week’s list of data news highlights covers January 27 – February 2, 2018, and includes articles about an AI system that developed a hit show in China and the largest genetic study ever.

1. AI Can Tell If a Baby’s Brain Is Healthy

Researchers at the Baby Brain Activity (BABA) Center in Helsinki have developed an AI system that can analyze electroencephalography (EEG) data about a premature baby’s brain activity and determine whether it is functioning properly. EEG data is useful for studying the brain activity of babies since doctors cannot perform traditional methods of testing brain function that rely on communication, however it can be difficult to meaningfully interpret EEG data. The AI system analyzes EEG data about a baby and compares activity patterns to historical data from other babies to determine if the baby’s brain activity appropriately reflects the baby’s age, as this activity changes as brains grow.

2. Making Roads Safer with Data Science

The U.S. Department of Transportation (DOT) has launched two pilot projects to use data analytics to improve highway safety. The first project will combine historical data about highway crashes, information about highway design, and anonymized GPS data indicating prevailing speeds across the entire national highway system in five minute intervals, so DOT can analyze how speed and road characteristics influence the risk of crashes. The second project will combine highway crash data with data from crowd-sourced traffic app Waze about road hazards and conditions to determine the feasibility of using Waze data to generate timely estimates of crash risks.

3. AI Picked China’s New Hot Show

Chinese video streaming company iQiyi, often referred to as China’s Netflix, created an AI system named iQiyi Brain that helped it develop hip-hop talent show Rap of China, which proved to be incredibly popular, receiving 2.68 billion views from its launch in July 2017 to the show’s end in September 2017. iQiyi Brain analyzes data from the 400 million users of the iQiyi app to identify factors that make shows and movies popular and can predict their performance up to a year in advance with 80 percent accuracy. iQiyi Brain predicted that even though hip-hop is not mainstream in China, Rap China would be successful, and also recommended the show’s producers use a particular celebrity as the show’s judge.

4. Predicting Patients’ Medical Outcomes

Researchers at Google have developed an AI system capable of predicting medical outcomes for patients as soon as they are admitted to a hospital, such as whether they will be discharged or readmitted, their diagnosis, and whether they will die in the hospital. The researchers trained their system on 46 billion de-identified data points generated by 216,221 adult patients from two hospitals in the United States over 11 years. The researchers claim their system can predict medical outcomes more accurately than current methods and patient deaths one to two days faster than current methods.

5. Australia Plans To Get More Innovative

Australian scientific advisory agency Innovation and Science Australia (ISA) has published a strategic roadmap for Australian innovation titled “Australia 2030: Prosperity Through Innovation,” which will guide government actions to promote innovation, with a particular focus on AI. The strategy provides 30 recommendations across five categories: providing the Australian workforce with the skills it will need to succeed by 2030; increasing productivity and stimulating high-growth industries; making government a world leader in innovation; promoting research and development; and spur innovation with large-scale challenges.

6. Understanding the Contents of Radiological Reports

Researchers at the Icahn School of Medicine at Mount Sinai have developed a natural language processing system that can analyze and extract useful information from the text of reports radiologists wrote about patients’ computed tomography (CT) scans. The researchers trained their system on 96,000 radiologist reports about head CT scans and taught it to identify relevant clinical words and phrases, such as “colonoscopy” or “phospholipid,” with an accuracy of 91 percent. The system could enable future AI systems to interpret large amounts of radiological texts to aid in diagnostics.

7. Rebuilding Puerto Rico with Smart City Technology

The government of Puerto Rico has partnered with the Smart Cities Council, a group of companies and research institutions developing smart city technology, to incorporate the Internet of Things as it rebuilds its infrastructure destroyed by Hurricane Marta. The Smart Cities Council will provide Puerto Rico with grant funding and free smart city technologies and services from participating companies. The partnership will also help the City of San Juan develop a smart city roadmap.

8. Fighting Traffic with AI

Alibaba has partnered with Kuala Lumpur to deploy an AI system called City Brain capable of increasing the city’s traffic speeds. City Brain analyzes video footage and data from transportation agencies, public transportation systems, and navigation apps to predict traffic patterns in real time, improve traffic flow, and detect traffic incidents. Alibaba has already deployed City Brain in Hangzhou, China and was able to increase average traffic speed by 15 percent.

9. Tracking Down the Genetic Cause of Insomnia

Geneticists at Vrije University in Amsterdam have conducted a study into the genetic roots of insomnia using genetic data from 1,310,010 people, making it the largest genetic study ever undertaken. The researchers analyzed data from consumer genomics company 23andMe and the UK Biobank, a public resource for large amounts of genetic data, and identified 956 different genes linked to insomnia, which could serve as targets for future research and potential treatments.

10. Predicting If a Flight Will Be Delayed

Google has developed a new feature for its Google Flights tool that uses machine learning to predict whether or not particular flights are likely to be delayed and notify users before the airline does. Google Flights will combine airline data with historical flight data to identify factors that make a flight likely to be delayed, such as location, weather, and other aircraft at an airport arriving late, and alert users when it is at least 80 percent confident their flights are delayed.

Image: Sham Hardy. 

You may also like

Show Buttons
Hide Buttons