Sepsis is one of the most dangerous medical challenges in America, the overreaction to infection killing millions of people and costing health care systems billions of dollars each year. Now hospitals and others have a chance to see if artificial intelligence will prove to be just what the doctor ordered.

The U.S. Food and Drug Administration this week gave De Novo approval to Prenosis for an AI program called Sepsis Immunoscore, a “revolutionary artificial intelligence diagnostic and predictive tool for early sepsis detection.”

The De Novo approval notes the novelty of the tool, since approval can’t be based on similarity to something else approved for marketing. According to the FDA, “The De Novo request provides a marketing pathway to classify novel medical devices for which general controls alone, or general and special controls, provide reasonable assurance of safety and effectiveness for the intended use, but for which there is no legally marketed predicate device.”

Running biomarker and clinical data through AI, the tool helps risk assessment for sepsis within a day of a patient’s arrival in the emergency department or hospital. It uses 22 different parameters to create a risk score, though it isn’t an alert system, Prenosis said in a news release. Instead, the risk categories generated by AI are related to the individual’s deterioration risk, what else is happening in the hospital and how care level needs increased in the first 24 hours.

“This combination of both diagnostic and predictive information had never previously been available in a legally marketed device for sepsis,” per the announcement.

About sepsis

In a given year, the Centers for Disease Control and Prevention reports that at least 1.7 million U.S. adults develop sepsis. And more than 350,000 people who develop sepsis in the hospital die as a result or leave the hospital to go into hospice care. The CDC said that a third of people who die in the hospital had sepsis while there.

Sepsis is not necessarily acquired in the hospital, but can lead to hospitalization. “Infections that lead to sepsis most often start in the lung, urinary tract, skin or gastrointestinal tract. Without timely treatment, sepsis can rapidly lead to tissue damage, organ failure and death,” the CDC said.

How Intermountain Health is working to improve outcomes for critically ill patients

Sepsis is not contagious, though infections that can cause it often are contagious. Most often, people with sepsis had one or more underlying medical conditions, per the CDC.

Symptoms include fast or weak pulse, confusion, shortness of breath, being feverish or very cold, extreme discomfort and clammy or sweaty skin. Sepsis is a medical emergency that puts certain people at increased risk, including those who are 65 and older, those who are immunocompromised, very young children and those who were recently hospitalized or severely ill.

Recruiting AI

The Prenosis tool is the brainchild of Bobby Reddy Jr., an electrical engineer who is CEO of Prenosis. As The Washington Post reported, he had seen “how patient care began with a diagnosis and followed a set protocol.” He believed an AI tool that looked at people individually would be more effective.

“In hospitals and emergency departments, we are still relying on one-size-fits-all, when instead we should be treating each person based on their individual biology,” he told The Post.

Sepsis can be deadly and occurs when the immune system ramps up to fight infection, but overreaches and attacks the body’s organs.

According to the release, Prenosis took more than 100,000 blood samples and information on hospital patients and trained its AI algorithms on the factors most often linked to sepsis. Then it narrowed that to the 22 parameters that made up its risk score.

Johns Hopkins University has been working on AI and sepsis, as well, as Johns Hopkins Magazine recently reported. “The team partnered with the two largest electronic health record system providers, Epic and Cerner, to ensure that the tool can be implemented at other hospitals,” the article said.

That team is also looking at using AI technology to spot patients who might get bed sores or see “sudden deterioration caused by bleeding, acute respiratory failure and cardiac arrest.”