Oct 04 2019

How Hospitals Are Using AI to Detect and Treat Sepsis

The technology is finding wider adoption in the fight against the deadly condition.

Responsible for more than 270,000 annual deaths in the U.S., sepsis claims a life in this country every two minutes. The condition, which arises from the body’s inflammatory response to infection, costs over $27 billion in hospitalizations each year. 

Despite advancements in understanding and managing sepsis, the fight is far from over. This is why an evolved strategy using predictive technology is critical.

By leveraging patient data, artificial intelligence is helping healthcare organizations identify patients in the early stages of sepsis. With the help of machine learning, custom dashboards to display risk scores and automatic alerts that notify caregivers of potential trouble, an AI-guided approach allows clinicians to get in front of the condition and even predict an adverse event.

Since introducing a sepsis warning system in 2017, Fishersville, Va.-based Augusta Health has witnessed a decrease in mortality rates from sepsis, saving an estimated 282 lives as a result. This drop in mortality rates is promising news and could prompt wider deployment of the tools.

Augusta’s warning system is able to examine patients’ vital signs via their electronic health records, recognize familiar warning signs of sepsis onset and alert clinicians and staff if abnormalities arise.

A number of healthcare organizations are doing the same by adopting or building upon existing AI models to help combat sepsis. Here’s a look at some notable recent efforts.

MORE FROM HEALTHTECH: Find out how AI is helping predict and prevent senior falls.

Duke University Hospital Builds an AI Sepsis Monitoring System

Sepsis can swiftly progress to organ failure and eventually to septic shock, which has a mortality rate of roughly 50 percent

It’s one reason the Duke Institute for Health Innovation has developed Sepsis Watch, an AI system that identifies early symptoms of sepsis. The system, which launched at Duke University Hospital in November 2018, is trained via deep learning and analyzes over 32 million data points to evaluate a patient’s condition in real time — and, if findings dictate, alert the hospital’s rapid response team. 

Response teams must make the call on executing next steps but they don’t have to do so alone. Sepsis Watch can guide nurses through the first three hours of care administration, walking them through a checklist of recommended treatment steps ranging from blood tests to medications. 

The pilot phase of the system was completed in May of this year and was successful. Since then, the healthcare system has expressed intent to expand the tool to other care settings within Duke’s hospital system and eventually to other hospitals.

READ MORE: Where does AI automation fit into health data security?

Sentara Healthcare Expands Data Points to Improve AI Model

In an effort to stop patients from contracting sepsis, Sentara Healthcare recently deployed a new AI system to analyze roughly 4,500 pieces of electronic health record data. The tool collects a patient’s EHR data — with details such as heart rate, blood test results and body temperature — and puts it through an algorithm to assess the likelihood of developing the condition.

Sentara’s new detection system closely examines the relationship between data points to more accurately predict future outcomes. 

When a patient is deemed high risk, the system is designed to alert the clinicians listed on the patient’s chart. It also suggests appropriate next steps, which could include more frequent monitoring of vital signs or cleaning a patient’s bronchial pathways to prevent pneumonia.

Earlier this year, the Norfolk, Va.-based healthcare system launched an initial attempt at a sepsis alert system that uses nine data points to alert doctors and nurses when a patient is at risk for the life-threatening condition. Dr. David Mohr, Sentara’s vice president of clinical informatics and transformation, tells Medical Xpress that the new system “goes way beyond” the previous iteration by notifying clinicians well before the patient develops the condition.

Despite the long-standing challenges of sepsis, the implementation and refinement of AI systems are offering clinicians a new level of support. Not only will this reduce healthcare costs, it will enable truly lifesaving care.

This article is part of HealthTech’s MonITor blog series. Please join the discussion on Twitter by using #WellnessIT.


monkeybusinessimages/Getty Images

Zero Trust–Ready?

Answer 3 questions on how your organization is implementing zero trust.