UF researchers can now assess and treat a patient’s condition faster than ever before with advanced artificial intelligence technology.
Dr. Azra Bihorac, a UF college of medicine and surgery professor, and Parisa Rashidi, a UF biomedical engineering assistant professor, published research last Tuesday in the Nature Scientific Reports journal that describes a system that speeds up and more accurately determines a patient’s condition using artificial intelligence.
The system Rashidi and Bihorac developed quickly processes a large amount of data at once in order to provide a clear interpretation of a patient’s medical condition.
“We do have the ability to help people not develop these devastating complications,” Bihorac said. “The only way we can improve outcomes is by preventing them using this technology until we develop new therapies one day.”
The research started in 2014 and is funded by a $2 million grant from the National Institute of Health and a $500,000 grant from the National Science Foundation, Rashidi said.
The system is called DeepSOFA, or Deep Sequential Organ Failure Assessment, and is more efficient than other models known as SOFA, or Sequential Organ Failure Assessment, Rashidi said. A SOFA score is a mortality prediction determined using set degrees of dysfunction in the six organ systems.
It gives a “score” to each organ to rank its functionality, Rashidi said. This traditional system is often used in intensive care units to predict in-hospital likelihood of death due to a patient’s condition.
In one patient’s case, the SOFA system originally predicted a five percent probability of death. Comparatively, the DeepSOFA system more accurately predicted a 50 to 80 percent probability of death, Rashidi said.
Essentially, the system is better at indicating when a patient is in need of an urgent life-saving procedure, Rashidi said.
The DeepSOFA can potentially improve patients’ outcomes who find themselves in intensive care or fatal conditions, Rashidi said. DeepSOFA lets doctors know exactly which patients need the most immediate medical attention first.
The algorithm developed by the team is patent pending by UF, Bihorac said. The technology has been tested for effectiveness using data from more than 85,000 prior patients at UF Health Shands Hospital and Beth Israel Deaconess Medical Center in Boston. To test effectiveness, each of the hospitals compared their own SOFA model data results with the DeepSOFA model.
“The next step in this is developing platforms and tools that can be used in real time,” Bihorac said. “In the future, we want our technology to be able to be applied at the bedside to assist physicians practicing medicine.”