10 Bits: the Data News Hotlist
This week’s list of data news highlights covers November 11-17, 2017, and includes articles about an AI fashion assistant for retail stores and the first approved smart drug in the United States.
The U.S. House of Representatives has unanimously passed the Foundations for Evidence-Based Policymaking (FEBP) Act, introduced by Speaker Paul Ryan (R-WI), which contains the Open, Public, Electronic, and Necessary (OPEN) Government Data Act—legislation that would codify federal open data publication requirements. The Senate had previously passed a version of the OPEN Government Data Act as an amendment to its version of the 2017 National Defense Authorization Act, but it was struck from the bill during conference committee.
Chinese e-commerce company Alibaba has developed an AI-powered fashion assistant called FashionAI that can help customers shop for clothes in retail stores. FashionAI relies on small sensors embedded in clothes and a screen in dressing rooms that can identify garments and recommend matching items. Alibaba customizes FashionAI for each store based on its inventory and has already deployed it in 13 stores across China.
Researchers at the University of California, San Francisco, and health startup Cardiogram have successfully demonstrated that the Apple Watch can determine if a user has high blood pressure or sleep apnea. The researchers used an AI system developed by Cardiogram called DeepHeart that analyzes data from the Apple Watch’s heart rate sensor to identify patterns indicative of high blood pressure and sleep apnea, such as lower heart rate variability. DeepHeart can predict whether or not users have these conditions with 80 percent and 90 percent accuracy, respectively.
Louisville, Kentucky, has recently completed a pilot study called Air Louisville that tracked the impact of Louisville’s notoriously high pollution levels on how 1,200 Louisville residents with respiratory conditions, such as asthma, used Internet-connected inhalers. The connected inhalers, provided by connected health startup Propeller Health, reminded users through their smartphones to take regular maintenance doses, resulting in an 84 percent decrease in emergency inhaler use by the end of the study. The pilot also shared data about each use with the city so it could identify factors contributing to air pollution hotspots and better target air quality improvement efforts.
Self-driving truck startup Embark has been delivering Frigidaire refrigerators to customers autonomously since early October. Embark, which partnered with trucking company Ryder for this pilot, uses human drivers in the cab to monitor the truck and take over in the event of an emergency during the course of the 650-mile trip from Frigidaire’s warehouse in El Paso, Texas, to the distribution center in Palm Springs, California. The trucks only drive themselves on the highway, pulling into rest stops to switch the trailer to a human-driven truck for the first and last legs of each trip, though Embark hopes to eventually completely automate this process.
The U.S. Food and Drug Administration has approved the world’s first microchip-containing drug, which can communicate with a smartphone once swallowed to verify that patients are adhering to their medication regimens. The drug is a new version of an existing antipsychotic medication called Abilify that can detect when it is submerged in stomach acid and transmit a signal to an adhesive patch on a patient’s torso, which then transmits data about dosage and ingestion time to a smartphone app that patients can monitor and share with caretakers. The smart pill is particularly useful for patients with mental illnesses, such as schizophrenia and bipolar disorder, as failure to adhere to medication regimens can make treatment less successful.
China now leads the world in supercomputing power after beating out the United States for having the most supercomputers on the Top500, a twice-yearly ranking of supercomputers around the world. China has had the two most powerful supercomputers for two years, but the United States had more supercomputers in the Top500 with a greater cumulative performance. In the six months since the last Top500 ranking, China has surpassed the United States in both the greatest number of of supercomputers on the list, increasing from 159 to 202, and total supercomputer performance.
The UK Biobank, an initiative supported by a host of medical charities, researcher institutes, and health agencies in the United Kingdom, published a massive freely-available database of genetic data on 500,000 British people in July 2017, spurring a wave of new genetic discoveries. For example, in October, biologists at King’s College, London, completed the largest ever genetic study of anxiety thanks to UK Biobank data. Additionally, British pharmaceutical company GSK is using the data to determine every potential health effect that blocking or boosting an individual gene may have to design better personalized treatments.
Belong, a recruitment startup based in Bengaluru, India, has developed an AI platform to recommend potential hires for its clients and rank them based on their suitability for a particular role. The platform aggregates publicly-available data about candidates, such as their LinkedIn and GitHub pages, to make recommendations, and automatically fine tune its rankings based on feedback from companies, such as whether or not they reject a particular candidate. Belong’s clients have reported that using the platform has improved the candidate selection process, boosting offer acceptance rates by 65 percent for one company and increasing candidate responses by 55 percent for another.
A group of astronomers have developed an AI system that can rapidly analyze satellite and observatory imagery of space to accelerate research into the composition of the universe. In space, when a large object comes between a distant light source and an observer on Earth, it distorts the space and light around it, allowing a glimpse of things that would normally be obscured by that object—a phenomenon known as a gravitational lense. The best human researchers can sort through images of space at a rate of 1,000 per hour to identify if an image contains a gravitational lense. In a test, the AI system operating on dated hardware was able to sort through 21,789 images in just 20 minutes, identifying on average one gravitational lense per minute. By contrast, the scientific community has only been able to identify approximately 100 gravitational lenses in the past several decades.
Image: Chris Watson.