Spanish artificial intelligence manages to predict the risk of sepsis 24 hours in advance.
A artificial intelligence developed in Spain achieve predict the risk of sepsis 24 hours in advance, that is, it predicts whether the body’s immune system will overreact to an infection, damaging its own tissues and organs.
Sepsis is a leading cause of death in hospitals as the infection causes a severe chain reaction throughout the body. So this AI application will allow reduce both mortality and morbidity associated with this condition.
The tool was developed by the Institute of Knowledge Engineering (IIC) and has has proven its effectiveness for more than two years at the Son Llatzer University Hospital in Palma de Mallorca. For this reason, it is being extended to other Spanish medical centers.
In particular, 170 Spanish hospitals have implemented this system to try to anticipate the occurrence of sepsis. But this system has a large number of false positives and negatives, so Detection speed varies greatly and ranges from 20% to 80%, depending on the experience of the medical team. For example, sometimes it is not detected until the first organic disorders appear or it is already too late.
It’s because Detecting sepsis ‘extremely difficult’because it can manifest itself as peritonitis, pneumonia, meningitis… It depends on a variety of conditions, as well as on the different genetic predisposition of the patient,” explained Marcio Borges, coordinator of the Sepsis Unit at Sean Liatzer Hospital.
“We looked at data from 9,300 patients with sepsis and 78,000 without sepsis. This allowed us to compare our machine learning model with traditional systems and with our own automatic detection program and see the differences in false positives. In the case of AI there were less than 9% false positives, the system was very sensitive and precise– Borges clarified.
“AI sees what we don’t see”
The goal was to find out whether the algorithm was able to tell the medical team in real time whether patients would have sepsis or not. It was carried out from January 2019 to April 2022 – during this period 64,524 sepsis alerts were issued, of which 30% were addressed. The tool showed that only 1% are false negativesthat is, the warnings are accurate and are unlikely to fail.
In short, Dr. Borges sums it up: “AI can work with large amounts of data in deeper path and sees what we don’t see,” with a predictive ability of 96%.
Given the success, two other hospitals, 12 October in Madrid and Complejo Hospitalario Universitario de Albacete, are implementing artificial intelligence software in their patient management systems. And the forecast is like this spread to other centerseven Europeans, given that this is the first tool of its kind to be successfully tested in the European Union.
Follow us on our WhatsApp channel and don’t miss the latest news and updates from anten3noticias.com.