Widely used AI tool for early detection of sepsis could dispel doctors’ doubts

Posted by

Proprietary artificial intelligence software designed as an early warning system for sepsis can’t differentiate between high- and low-risk patients before they receive treatment, according to a new study from the University of Michigan.

Photomicrograph of S. aureus bacteria.

Photomicrograph of S. aureus bacteria. Image Credit: Dr. Richard Facklum, USCDCP

The tool, called the Epic Sepsis Model, is part of Epic’s electronic medical record software, which serves 54% of patients in the United States and 2.5% of patients internationally, according to a statement from the company’s CEO. informed of By Wisconsin State Journal. It automatically generates sepsis risk estimates in hospitalized patients’ records every 20 minutes, hoping to allow physicians to detect when a patient may have sepsis before things get worse.

“Sepsis has all these vague symptoms, so when a patient comes in with an infection, it can be really difficult to know who can be sent home with some antibiotics and who needs to stay in the intensive care unit.” Can. We still miss a lot of sepsis patients,” said tom valleyAssociate professor in pulmonary and critical care medicine, ICU physician and co-author of a study recently published in the New England Journal of Medicine AI.

Sepsis is responsible for one-third of all hospital deaths in the US, and prompt treatment is key to patient survival. The hope is that AI predictions can help do this, but currently, they are not able to get more out of patient data than physicians can.

“We suspect that some of the health data that the Epic Sepsis model relies on encodes, perhaps inadvertently, the physician suspecting that the patient has sepsis,” it said. Jenna WiensAssociate professor of computer science and engineering and corresponding author of the study.

For example, patients may not receive blood culture tests and antibiotic treatment until they begin showing symptoms of sepsis. While such data could help AI very accurately identify sepsis risks, it may enter the medical record too late to help physicians move forward with treatment.

This mismatch in the time between when information is available to the AI ​​and when it is most relevant to physicians was evident in the researchers’ evaluation of the Epic Sepsis Model at the University of Michigan Health, the clinical branch of Michigan Medicine. How it fared for the 77,000 adults hospitalized. ,

The AI ​​already predicted each patient’s risk of developing sepsis in the medical center’s standard operations, so all the researchers had to do was extract the data and analyze it. About 5% of patients had sepsis.

To measure the AI’s performance, the team calculated the probability that the AI ​​gave higher risk scores to patients who were diagnosed with sepsis, compared to patients who were never diagnosed with sepsis.

When incorporating predictions made by AI at all stages of a patient’s hospital stay, AI can correctly identify a high-risk patient 87% of the time. However, the AI ​​was correct in only 62% of cases when using patient data recorded before they met the criteria for having sepsis. Perhaps most telling, the model gave high risk scores to only 53% of patients who had sepsis when predictions were restricted to before blood cultures were ordered.

The findings showed that the model was taking into account whether patients received a diagnostic test or treatment when making predictions. At that point, physicians already suspect that their patients have sepsis, so AI predictions are unlikely to make any difference.

Donna Tjandra, a doctoral student in computer science and engineering and co-author of the study, said, “We need to consider when the model is being evaluated in the clinical workflow when deciding whether it will help physicians. Is it helpful or not?” “Evaluating the model with data collected after the physician already suspects sepsis onset may make the model’s performance appear stronger, but it does not align with assisting physicians in practice.”

Source: University of Michigan


#Widely #tool #early #detection #sepsis #dispel #doctors #doubts

Leave a Reply

Your email address will not be published. Required fields are marked *