A new high-performance contact lens (隐形眼镜) under development at the University of Heidelberg will not only correct ordinary vision faults but will improve normal night vision as much as five times, making people’s vision sharper than that of cats.

Bille and his team work with an instrument called an active mirror. Connected to a wave-front sensor that tracks and measures the course of a laser beam (激光束) into the eye, the aluminum mirror discovers the faults of the cornea (眼角膜) , the transparent protective layer covering the eyeball. The highly precise data from the two instruments, which, Bille hopes, will one day be found at the opticians (配镜师) all over the world — serve as a basis for the production of completely individualized contact lenses that correct and improve the wearer’s vision.

By day, Bille’s contact lenses will focus rays of light so accurately on the retina (视网膜). Lens wearers will be able to identify a face at a distance of 100 meters — 80 meters farther than they would normally be able to see. At night, the lenses have an even greater potential. Normally, people can see a face at a distance of 5 metres. In his experiments night vision was improved by an even greater factor: in semi-darkness, test participants could see up to 15 times better than without the lenses.

Bille’s lenses are expected to reach the market soon, and one tentative plan is to use the Internet to transmit information on patients’ visual faults from the optician to the manufacturer, who will then produce and mail the contact lenses within a couple of days. The physicist expects the lenses to cost about a dollar a pair: about the same as conventional one-day disposable lenses.

When my 8-month-old cries, I ask him if he’s hungry, or wet or just needs a hug.

“Babububuu,” he says.

What I need is a baby cry translator. That’s just what a team of researchers say they’ve developed.

“Experienced nurses or pediatricians (儿科医师) can identify why baby is crying because they have experience, says Lichuan Liu, a professor of electrical engineering at Northern Illinois University, who conducted the research.” We talked to them, and they mentioned that based on the cry’s’ sound there’re’ some clues (线索).”

So Liu set out to identify the features of cries that can help mark them as expressions of pain or discomfort. These features include differences in pitch (音高) and frequency. The team then developed an algorithm (算法) based on automatic speech recognition to detect and identify these features. This “cry language recognition algorithm” was trained on recordings of baby cries taken from a hospital. It uses compressed sensing, a process that reconstructs a signal based on incomplete data. It can identify a baby cry against a background of, say, adult speech or loud television sounds. By classifying different cry features, like pitch, the algorithm can suggest whether the cry is due to sickness or pain, and identify the degree of urgency.

The team had experienced pediatric care providers assess forty-eight baby cry recordings for probable cause of crying hunger, tiredness, etc. They then compared these to the algorithm’s assessments. The algorithm agreed with the humans 70% of the time. Generally, so-called “uncommon cry signals”—signs of pain or sickness—are high-pitched and very loud compared to ordinary crying.

Liu and her team continue to train the technology for greater accuracy. They also plan to add more features, like the ability to identify and classify movement and facial expressions. This could help give more detailed reading of baby’s emotional (情感的) and physical state. They also hope to begin human trials in the near future.

Scientists have developed an AI system focused on translating a person’s brain activity into a stream of text, according to a study published Monday in the journal Nature. The system, called a semantic decoder (解码器), could finally benefit patients who have lost their ability to physically communicate after suffering from a stroke or other diseases.

Researchers at the University of Texas at Austin developed the system in part by using a model, which is similar to those of Google’s chatbot Bard and OpenAI’s chatbot ChatGPT. The study’s participants (参加者) trained the decoder by listening to several hours of podcasts within an FMRI scanner (扫描仪), which is a large machine that measures brain activity. Once the AI system is trained, it can make a stream of text when the participant is listening to or imagines telling a new story. The text is not an exact copy, but the researchers designed it by catching general thoughts or ideas.

According to a news release, the trained system produces text that closely matches the meaning of the participant’s words around half of the time. For example, when a participant heard the words “I don’t have my driver’s permit yet” during an experiment, the thoughts were translated to, “She has not even started to learn to drive yet”.

“This is a real improvement compared to the system before, which is only single words or short sentences,” Alexander Huth, one of the leaders of the study, said. “We’re getting the model to decode continuous language with rich ideas.” Participants were also asked to watch four videos without sound while in the scanner, and the AI system was able to exactly describe “certain events” from them, the news said.