This week’s list of data news highlights covers February 23-March 1, 2019, and includes articles about an AI chatbot fact checking conversations on messaging apps and an AI system that predicts the power output of wind turbines.
Researchers from the Georgia Institute of Technology and Newcastle University in the UK have developed an AI system that identifies infants displaying symptoms of a perinatal stroke, which can cause life-long disabilities such as cerebral palsy. The system analyzes data from accelerometers wrapped around infants’ ankles and wrists to detect abnormal body movements. During a trial involving 34 infants, the AI system correctly identified 80 percent of the likely cases of a perinatal stroke.
Researchers from the University of California, San Francisco, and Baycrest Health Sciences, a research hospital in Toronto, are conducting a 100-person trial to test if ElliQ, a small robot companion, can alleviate feelings of loneliness in older adults. The robot, which Israeli technology company Intuition Robotics created, recognizes faces, communicates spontaneously, and can lean in to demonstrate an interest in a conversation. ElliQ also can prompt relatives to start video chats with its owner.
Asian companies are using AI to track the health of pigs, many of which are dying from African swine fever. For example, Alibaba has created voice recognition technology to monitor pigs’ coughs, and SmartAHC, an agri-tech startup based in Singapore, has created AI technology that tracks pigs’ body temperatures and exercise. In addition, Chinese e-commerce company JD.com has created facial recognition technology to detect if pigs are sick.
Researchers from Huawei and the Polytechnic University of Catalan in Barcelona have created an AI system that improves the efficiency of optical telecommunications networks. The researchers used reinforcement learning and deep learning to train the virtual agent on 5,000 simulations, teaching the agent how to optimize space in the network based on different types of traffic, such as a voice call or a streaming video. Increasing the efficiency of optical networks could reduce latency and costs in telecommunication.
Google and DeepMind are using machine learning to predict the power output of Google’s wind turbines 36 hours in advance. Using a neural network trained on weather and historical turbine data, their model recommends how to make optimal hourly delivery commitments to the power grid. The model has increased the value of Google’s wind energy by 20 percent.
Taiwanese developers have created Meiyu, an AI fact-checking chatbot that interjects in conversations on messaging apps to point out factual errors and provide alternative interpretations. More than 110,000 users of the Japanese messaging app Line have added Meiyu to their chats.
Researchers from Ben-Gurion University in Israel and the University of Texas Southwestern Medical Center have developed a method that uses AI to predict the spread of melanoma. The researchers used microscopic cameras to film cancer cells and AI to analyze the appearance and behavioral patterns of cells. This analysis allowed the researchers to predict the likelihood that stage III melanoma will progress to stage IV, a stage in which the cancer has spread to other areas of the body.
Japanese technology company NEC and Taiwanese Bank E.Sun have installed facial recognition-enabled ATMs in Taiwan. The ATMs use NEC’s facial recognition technology called NeoFace to enable customers to use their face to withdraw cash after a brief setup at one of the ATMs.
Researchers from the University of Toronto have used machine learning to predict how arthritis will progress in children. The researcher’s algorithm analyzed data from 640 children and sorted the disease into seven distinct types based on the location of swollen or painful joints. The algorithm then accurately predicted which children would develop a more severe form of arthritis.
Chinese hospitals are using facial recognition technology to identify individuals who illegally sell hospital appointments. In China, patients must visit public hospitals on the day they wish to see a doctor, allowing scalpers to wait in line to schedule an appointment they plan on selling. Facial recognition has helped the hospitals identify more than 2,100 scalpers.