Estimating Sepsis Risk
Sepsis is a common, life-threatening condition that results from systemic infection. It often starts with a local infection that’s undiagnosed or seems innocuous until it suddenly begins to rapidly spread through the body, triggering massive inflammation, damaging tissue, and shutting down vital organs, causing premature death or long-term disability.
Sepsis can occur at any time in patients of any age. More than a million Americans develop sepsis each year and 250,000 die from it. This article in the New England Journal of Medicine states that the estimate of the annual worldwide toll due to sepsis—30 million cases and 6 million deaths worldwide—is likely significantly low.
The longer it takes to identify and begin treating sepsis, the more deadly it’s likely to be. To improve outcomes, physicians and other healthcare providers need the ability to recognize early signs and quickly begin management, treat the infection, and support patients with the appropriate modalities.
Sepsis can manifest itself through a wide variety of vectors—everything from insect bites to urinary tract infections to flu-like symptoms. Sometimes healthcare providers don’t see the patients—or recognize the signs of sepsis—until organ failure is well established. Quickly identifying patients who have developed sepsis—or who are in danger of developing it—is key to early aggressive management and improved outcomes.
There’s a lot to consider, including which clinical and laboratory data indicate subtle and overt organ dysfunction, how to control infection at the source, and protocols for quickly recognizing early signs of septic shock. For example, providers aren’t always sure about when to start antibiotics, the importance of reassessing lactate levels, or which vasopressor to use for septic shock.
Doctors in ICUs and emergency rooms often use the Sequential Organ Failure Assessment (SOFA) score to assess patients’ sepsis risk. The score is based on individual scores for six systems in the body.
AI is a different, powerful tool for analyzing data to predict the likelihood that a patient has developed—or will develop—sepsis. Machine learning models built using DataRobot can extract deep insights from demographic and historical data from patients who developed sepsis in the past. Examining patterns in data related to symptoms, previous diagnoses, health background, and admission history—the models can predict which patients currently under care are at risk, with demonstrable results that can easily be explained to colleagues. This enables medical providers to start immediate, aggressive, customized treatment, saving precious time and lives.
Predictions made using DataRobot’s models are intended to supplement diagnoses made by physicians and other medical providers—and are in no way meant to replace their medical expertise.
DataRobot is one of the most widely deployed and proven AI platforms in the market today, working to deliver value to all industries, including healthcare. Its proven impact has attracted healthcare providers across the world. Machine learning models built using DataRobot have helped hospitals improve patient outcomes, reduce costs, and maximize revenue.
DataRobot AI Cloud platform is an ideal tool for assessing the risk of sepsis in a patient population, improving prevention, and avoiding severe outcomes. Its models are designed to analyze historical data of different types from multiple sources, identifying patterns, and delivering quick results. Models can be retrained as new data comes in, ensuring results reflect the latest findings and protocols.
DataRobot explains its predictions by highlighting the most influential factors so results can be demonstrated to patients, caregivers, executives, and regulators. And security and governance are built in.