This week’s list of data news highlights covers October 13-19, 2018, and includes articles about an AI chatbot that can reduce unnecessary doctor’s visits and a smartphone app helping disaster response efforts in Florida.
1. Using Smartphones to Measure Mental Health
A startup called Mindstrong Health has created a smartphone app that tracks how individuals use their phone to measure their cognitive and emotional health. Researchers have demonstrated that various brain disorders can influence how a person uses their phone, such as how fast they scroll or sort through contacts. Mindstrong’s app tracks activity such as how a person types, taps, and swipes and has compiled five years’ worth of clinical study data. It is now working with mental health researchers to use machine learning to uncover relationships between phone use behavior and mental health.
2. Developing Autonomous Ships
Rolls Royce is developing autonomous ships that use AI, and a combination of cameras, LIDAR, radar, and other sensors, to navigate waterways. Rolls Royce is conducting a trial of its ships in Norway, where autonomous ferries handle some sailing tasks while a human is still on the ship. The potential benefits of autonomous shipping include more efficient vessels because they do not need amenities such as heating and fewer crashes—at least 70 percent of the 1,000 large ship losses in the past decade were due to human error.
3. AI is Helping Reduce Visits to the Doctor
Babylon Health, a healthcare provider in the United Kingdom, has created an app to allow users to seek medical advice from an AI-powered chatbot. The app uses natural language processing and machine learning to understand patients’ symptoms and then links the symptoms to conditions. The app has resulted in half of users opting not to seek treatment as a result of the chatbot providing useful information, however the app will always recommend seeking a second opinion from a human.
4. Racing Autonomous Cars
An autonomous car racing competition called Roborace will begin its first season in 2019. The races will consist of two laps around a racetrack—one with a human in control, and another with AI driving. The team with the fastest average human and AI driver times will be declared the winner. In a previous autonomous car race at the Berlin ePrix, an annual race in Germany, one AI driver was only a tenth of second slower than its human driver.
5. Seismic Data Could Predict How the Earth’s Ice is Melting
Scientists at Colorado State University studying data about seismic waves in Antarctic glaciers have discovered that glaciers emit a hum in frequencies too low for humans to hear. The waves shift based on events such as heavy storms and changes in temperature, and the scientists discovered that changes in a glacier’s hum could reflect how ice shelves are responding to warming temperatures and other weather patterns.
6. AI is Helping Train Cybersecurity Students
Texas A&M University is overcoming a cybersecurity workforce shortage by pairing 10 cybersecurity students with AI systems to detect and remediate the roughly one million cyber attacks it receives each month. An AI system helps students without cybersecurity experience identify significant potential threats that could warrant additional investigation. The university’s security operations center has full-time employees to monitor cyber threats, but now relies on 10 students using its AI software as the majority of its security staff.
7. Helping Doctors Make Predictions During Surgery
University of Washington researchers have developed a machine-learning system called Prescience that can predict if a patient will develop hypoxemia, a condition where blood oxygen levels go below normal, while under anesthesia. The researchers trained their system on data from 50,000 surgeries, allowing it to use data from patient health records and operating room sensors to predict patient hypoxemia risk before and during surgery. The system also provides explanations, such as a patient’s body mass index, for why an individual might be at higher risk.
8. Boosting Natural Disaster Recovery with a Smartphone App
People participating in the recovery effort for Hurricane Michael in the Florida Panhandle are using an app called Crowdsource Rescue (CSR) that allows people in distress to share their location so they can receive aid. Two developers created CSR during Hurricane Harvey as a web service, and by the time the storm was over, CSR helped 25,000 people recieve aid. CSR aggregates calls for help from social media, plots their geolocation data, and routes volunteers to the people in need.
9. A Quantum Computer Just Did Something a Classical Computer Never Could
Researchers from IBM, the University of Waterloo, and the Technical University of Munich have successfully demonstrated for the first time that a quantum computer could perform an operation not possible on a classical computer. Quantum computing researchers have developed algorithms that a classical computer could never run but that a quantum computer could in theory, but have never been able to prove it in practice due to the nascent nature of quantum computing. The researchers built a quantum circuit that computes a complex formula that requires something to be in two places at once—an impossible feat using classical computing circuits.
10. Making Drug Discovery Collaborative with AI
Researchers at the Massachusetts Institute of Technology have developed an AI system that can divide up a pool of pharmaceutical research data across multiple servers to train neural networks on the whole dataset while keeping the data encrypted. The system encrypts the data distributed across multiple servers to prevent misuse, while still allowing researchers to train other AI systems on this data to predict novel drug-protein interactions. This approach could encourage healthcare providers, academics, and others with large amounts of health data to share this data to advance research without having to worry that it could be stolen or misused.
Image: NASA.
