What telltale features—many inaudible to the human ear—separate one kind of cough from another? Scientists are on the verge of finding out with a new machine learning tool aimed at identifying the signature sounds of tuberculosis.

Tuberculosis is a major public health concern, an ancient bacterial disease that has claimed the lives of kings, presidents, poets and at least one star of Hollywood's silver screen-era. Yet even now in the 21st century, it's still impossible to shake the scourge. TB kills someone around the globe every 22 seconds, the World Health Organization estimates.

In clinical trials, a time-honored but old-school way to determine if TB is being knocked out by antibiotics involves having study participants cough up phlegm for a sputum culture, a test that can gauge whether the bacteria are succumbing to—or resisting—treatment.

Inadequate development of new antibiotics and rising rates of resistance by bacteria to existing antimicrobials are dual forces pushing the world ever closer to a post-antibiotic era.

The bacterial pathogen that causes tuberculosis is a master of deception, a king of clever tricks—an enemy agent that not only infiltrates but can become a long-term stowaway in patients' lungs.