This week’s list of top data news highlights covers September 28, 2024 to October 4, 2024, and includes articles on using quantum machine learning algorithms to detect lung cancer and using generative AI to create more engaging ads.
1. Customizing Children’s Reading
Ello, a San Francisco-based company that provides an AI companion to help children practice and improve their reading skills, has launched a new tool called Storytime that makes reading practice more engaging. Storytime allows children to participate in creating the stories they read in an app or on the Ello website by generating personalized stories using AI from settings, characters, and plots that children can choose from based on their interests.
2. Improving Argumentation Skills
A law professor at the University of Missouri is using an AI chatbot to help her students improve their argumentation skills and prepare for real-world scenarios. The chatbot, which the professor trained to be aggressive and challenging, acts as opposing counsel in mock trial simulations to challenge students on obtaining documents in a given case—a likely scenario for many lawyers in a trial case.
Researchers at the Indiana University School of Medicine have launched a tool called RelateXR that uses virtual reality (VR) to help fight drug addiction by showing recovering addicts what their future without drugs could look like. RelateXR creates custom avatars from user images and uses cloned versions of their voices to immerse them in a scenario 15 years into the future. Once the user puts their VR headset on, they can meet two possible future selves—one in recovery, and one still using drugs—allowing them to see the benefits of recovery more closely.
Researchers at Nirma University in India have published a study demonstrating how quantum machine learning (QML) models, which use quantum computing to speed up developing machine learning tools, can improve early lung cancer detection. The researchers used two QML algorithms to analyze a dataset that included 26 health-related attributes from roughly 1,000 patients. One of the algorithms reached 85 percent accuracy in detecting lung cancer patients, which demonstrates the potential of QML algorithms in making early detecting less costly and faster than current methods allow.
Numa, a California-based startup providing AI tools for car dealerships, is helping dealerships automate tasks like following up with missed calls, booking appointments, and providing service updates using AI. The tool helps improve the customer service experience at car dealerships as many businesses struggle to respond to all customer requests on time.
Microsoft has added features that use AI to improve Windows search on the company’s new Copilot Plus PCs. The new features will allow users to find images, files, and documents more quickly by using AI to analyze specific keywords that users input.
7. Producing Accurate Health Insights
Oura, a Finland-based health technology company, has upgraded the company’s smart ring with a new algorithm that allows the ring’s sensors to pull data from the most optimal place on the ring surface by increasing the number of signal pathways that the smart sensors use to capture data at a given time. The new algorithm increases signal accuracy by 120 percent and accuracy in tracking a user’s blood oxygen by 30 percent.
Pinterest has launched new tools for advertisers that help them to transform plain product images into lifestyle scenes to enhance engagement with ads on the platform using generative AI. The new tools help advertisers quickly create visually appealing ads, reduce input time, and boost click rates.
9. Predicting Fatal Conditions
U.S. Army researchers at the Walter Reed Army Institute of Research have developed SeptiBurnAlert, a system that predicts sepsis in burn patients—a potentially fatal condition when the body responds improperly to infection—using machine learning. After nurses use traditional hospital equipment to collect a patient’s blood, SeptiBurnAlert’s machine learning algorithm analyzes the data to forecast sepsis onset and alerts doctors days in advance.
Researchers at UCLA have developed a deep learning tool that reduces the time and cost for analyzing 3D medical scans to predict disease risk factors. Unlike standard 2D medical images, 3D images allow for further insights into the organ or disease but require significantly more effort and time to analyze. The tool matched clinical expert accuracy across multiple imaging modalities like MRI and CT scans while reducing annotation time.
Image credit: Dieter Blom