The latest in computer-based imaging could finally reveal the ancient Coptic writing inside a charred, 1,500-year-old parchment codex containing the Acts of the Apostles.

The manuscript, which the Morgan Library & Museum in New York obtained in 1962, has since remained unopened for fear destroying its fragile pages. The pages are thought to have been fused together by a cinder that sank through the document, causing the parchment fibers to congeal, according to The New York Times.

W. Brent Seales, a computer science professor at the University of Kentucky, spent 14 years developing a computer algorithm designed to read digital images of ancient scrolls deemed too fragile to unwrap. In partnership with Paul Dilley, a Coptic scholar at the University of Iowa and an expert on early Christianity, Seales agreed to tackle the Morgan Library codex—code-named M.910. He brought scanning equipment to the library because the codex was too delicate to travel.

Although detailed CT scans can reveal the ink letters inside a scroll or a codex (a book-style manuscript), the text is unreadable unless each letter can be assigned to the appropriate place on the surface of the document. Professor Seales developed software capable of modeling the undulating surfaces of ancient parchments or papyrus based on X-ray images and then assigning the letters to their proper place, according to the Times. Although Seales had never worked with a codex before—with writing on both sides of the pages—initial tests on a parchment mock-up showed the technique could work.

The scans on M.910 were completed in December, and Seale’s office at the University of Kentucky says that readable pages of the document should be available early this year.

The codex was likely written in Egypt between A.D. 400 and 600, before the Arab conquest in 642. The Acts portion of the document should be informative for showing the text’s consistency with earlier versions. Dilley hopes the results of the analysis will reveal what other writings may be found alongside the copy of Acts.

Devrimb/iStock

Emergency assessment

For 911 emergency dispatchers, recognizing the signs of cardiac arrest during a medical emergency call is critical. The dispatcher must quickly assess the symptoms being described by a caller who may be confused and frightened. Now an artificial intelligence assistant is helping emergency dispatchers in Copenhagen, Denmark, quickly and accurately diagnose cardiac arrest by listening in on 911 calls.

The technology, called Corti, analyzes verbal and nonverbal clues such as tone of voice, breathing patterns, and background sounds, comparing them to millions of other emergency calls Corti has previously analyzed. Corti then predicts the seriousness of the situation, providing alerts and recommendations to the dispatcher in real time.

A recent study found that dispatchers in Copenhagen recognized cardiac arrest from over-the-phone descriptions about 73 percent of the time, according to Fast Company. But Corti’s artificial intelligence system was correct 95 percent of the time. The accuracy of the system is also expected to continuously improve with each new call.

Corti’s developers say the tool is designed to work in the background to supplement, not replace, human dispatchers.