Every year more than a million Americans experience sepsis—a condition in which the immune system runs wild and damages its own tissues, sometimes with fatal results. Yet while sepsis is the leading cause of death in hospitals, few people outside of the medical profession recognize it by name.

“I think sepsis flies under the radar because it is elusive,” says Michael Filbin, an emergency medicine physician at Massachusetts General Hospital and one of the leaders of its sepsis initiatives. “It can be elusive even to health care professionals.” Sepsis is sparked by an infection, which means that it often lurks in the shadow of another disease. Its hallmarks are also subtle and easy to miss—part of the reason behind a nationwide push to come up with better protocols to help spot it.

One solution may be to enlist artificial intelligence to catch signs that physicians might miss. “We’re working with data scientists at MIT to develop machine-learning models that might help us detect sepsis earlier,” says Filbin. But AI has had limited success so far. Another hope is to develop better tests, including a new effort to use the latest techniques of genetic screening—a solution that is promising, but many years away.

What tools can help today? Listen here or subscribe to the Proto podcast on iTunes or Stitcher.