Experimental technologies to clean infected blood are raising hopes for new progress against an old killer: sepsis, a frighteningly common infection that kills more than 250,000 people a year in the United States alone.

Despite decades of research, sepsis still causes a third to a half of all deaths in US hospitals.

“In terms of human misery and death, this is way up there,’’ said Dr. Daniel Kohane, a professor of anesthesia at Harvard Medical School and the director of a lab working on a blood-cleaning device.

Sepsis starts with an infection — perhaps from a dirty surgical tool, a common scrape, or an illness such as pneumonia. The body releases an arsenal of chemicals to fight the infection; sometimes, they spark an inflammatory reaction that can be so severe it overwhelms the body. Sepsis can lead to fevers, difficulty breathing, and organ failure. It’s often so painful that the Centers for Disease Control and Prevention lists “I feel like I might die’’ as a common symptom.

One idea to help patients comes from Opsonix, a Cambridge startup launched last fall with $8 million in funding.

Its team has engineered a protein that can latch on to infectious particles and pull them out of the blood, said Eric Devroe, one of the company’s founders.

Opsonix plans to send patients’ blood through tubes coated on the inside with this protein. The infectious agents — bacteria, viruses, fungi, and other toxins that overwhelm the body during sepsis — would stick to the tubes. Newly cleansed, the blood would then be returned to the patient. It’s so far been tested only in animals.

Another company, Toray Industries, markets a similar device for clinical use in Japan and Europe, but it filters out fewer infectious agents, said Devroe. That device is not currently approved in the United States.

Researchers in Kohane’s lab, meanwhile, are trying a different approach, focused on the second stage of sepsis — when the immune system overreacts to a run-of-the-mill infection.

To combat it, Kohane’s lab is developing a blood filter that removes specific molecules, called cytokines, which are released by the immune system and whose elevated presence can harm the body. The researchers are also engineering a monitor to measure cytokine levels in near-real time.

Currently, it takes hours to measure the level of these molecules in the blood. The team hopes its monitor will be able to do it in real time. Once the scientists know which cytokines are present in the blood, they’ll be able to adjust the filter and remove those specific dangerous agents.

The team has not launched a company yet, but it recently nabbed $31,000 in prize money from two different competitions at the Massachusetts Institute of Technology. It also took home “bragging rights and an Apple Watch’’ as an award from an innovation summit hosted by Boston Children’s Hospital, said team member Dr. Brian McAlvin. He gave the watch to his wife.

“I got zip,’’ Kohane grumbled.

The group is working to set up a trial of the filtration device in pigs and continues to study it in rats.

Sepsis research is also advancing at a more basic level, as researchers learn to recognize it earlier in patients and to standardize treatment protocols. An article in the Journal of the American Medical Association last month set out to better define sepsis. And last year, the Centers for Medicare and Medicaid Services mandated that hospitals start collecting data on how fast sepsis is recognized and treated.

“This is an incredibly exciting time to be involved in sepsis research,’’ said Dr. Craig Coopersmith, a professor of surgery at the Emory University School of Medicine. “We understand more about the nature of sepsis than ever before.’’

Whether or not the new technologies pan out, deaths can be reduced with basic steps, such as educating doctors to do a blood culture and start the patient on broad-spectrum antibiotics as soon as sepsis is suspected, said Dr. Andreas Taenzer, an anesthesiologist at Dartmouth-Hitchcock Medical Center in New Hampshire.

Such protocols helped the hospital cut their sepsis deaths and costs in half.

Taenzer said he is intellectually intrigued by technologies like the blood filtering devices, but that the most important step is recognizing sepsis early.

“Without that,’’ he said, “nothing flies.’’

Ike Swetlitz can be reached at ike.swetlitz@statnews.com. Follow him on Twitter @IkeSwetlitz. Follow Stat on Twitter: @statnews.