Home BlogWeekly News 10 Bits: the Data News Hotlist

10 Bits: the Data News Hotlist

by Michael McLaughlin
Power plant creating a large white plume.

This week’s list of data news highlights covers May 4-10, 2019, and includes articles about an AI system tracking air pollution and the UK using AI to update maps in near real time.

1. Predicting a Person’s Likelihood of Developing Breast Cancer

Researchers from MIT and Massachusetts General Hospital (MGH) have developed an AI system that can predict if a patient is likely to develop breast cancer up to five years in advance. The researchers trained the system on mammograms and known outcomes, such as if a person developed breast cancer, of 60,000 MGH patients. The system could help doctors better determine mammogram schedules for women depending on their risk of developing breast cancer.

2. Tracking Air Pollution in Real Time

WattTime, an AI nonprofit, has developed an AI system that analyzes satellite imagery to track the air pollution of all power plants in the world in real time. The system analyzes images for visible smoke and infrared images for heat from smokestack plumes to detect emissions. Poor monitoring of power plants allows some actors to evade restrictions, and this system can help provide data that verifies if power plants are meeting requirements.

3. Automating Taste Testing
China has developed taste-testing AI robots as part of a government-funded program to automate the tasting of mass-produced food, including vinegar. The robots, which use sensors to simulate human eyes, noses, and tongues and a neural network to analyze the data, automatically adjusts the conditions of production, such as pace, to ensure food has the correct color, smell, and taste. The robots finish testing in less than a second and are almost as accurate as human tasters.

4. Taking Photos from 28 Miles Away

Researchers from the University of Science and Technology of China have developed a system that uses LIDAR and a computational imaging algorithm to create photographs up objects up to 28 miles away. The system uses LIDAR to illuminate and detect single photons while the algorithm knits together data points to create the image. Photons return to the system at time intervals in relation to their distance, allowing the algorithm to ignore any photons that arrive outside the desired window.

5. Predicting Children at Risk of Not Being Vaccinated

Researchers from the United States, France, Portugal and Croatia have developed a machine learning model that predicts children at risk of not being vaccinated with 72 percent accuracy. The researchers trained the model on the electronic health records of 48,000 children entering the first grade between 2011 and 2018. The model can help public officials and physicians identify and talk to the families of children with the highest risk of not receiving vaccinations before the families have decided they will not vaccinate their children.

6. Updating Maps of the UK in Near Real Time

Ordnance Survey, the UK’s national mapping agency, has begun a pilot project to create near-real-time maps of the nation’s streets by mounting cameras to the dashboards of utility companies’ vehicles. Algorithms process the data from the cameras, detecting road signs, traffic lights, drains, and other road and roadside details. The maps can help the UK more accurately and efficiently manage its infrastructure.

7. Automating the Detection of Creatures in the Seabed

Researchers from the University of Plymouth in the UK have shown that computer vision can help scientists identify the species living on the ocean floor. Biologists need information about the animals that inhabit the seabed to inform conservation efforts, but manually annotating images from autonomous underwater vehicles, which can capture more than 100,000 images in a single dive, can be slow. The researchers trained a convolutional neural network on 20 to 1,000 images per taxa living on the seabed, finding that the network could be up to 93 percent accurate depending on the type of animal.

8. Detecting the Early Signs of Lung Cancer

Google has developed an AI system that can detect the early signs of lung cancer. Google trained the system on lung cancer computed tomography (CT) scans from the National Cancer Institute and Northwestern University, allowing it to detect signs of cancer that are hard for oncologists to notice. In one case, the system identified the early signs of cancer in a patient who developed the disease despite five out of six radiologists missing the signs.  

9. Creating an AI Hummingbird

Researchers from Purdue University have developed an AI robot that can fly and maneuver in the manner of a hummingbird. The researchers created simulations of hummingbird flights to train the robot, teaching it how to perform maneuvers such as rapid 180-degree turns. The robots, which are as small as one gram, could help reach areas in rescue missions that humans can not access.

10. United States Congress Creates AI Task Force

The U.S. House of Representatives Committee on Financial Services has created the Task Force on Artificial Intelligence. Congressman Bill Foster (D-IL) will chair the task force, which will examine topics including the impact of automation on jobs in financial services, the role of algorithms and big data in risk management, and how the U.S. can use AI to maintain its competitiveness in the financial services sector

Image: Tony Webster

You may also like

Show Buttons
Hide Buttons