TECH

IA achieves better results than doctors predict death situations

An investigation presented at an international conference in Lisbon revealed that Artificial Intelligence (AI) systems are better at identifying death situations than human doctors. The research was led by Dr. Luis Eduardo Juarez-Orozco of the Turku PET Center in Finland, who counted as the training team a machine learning algorithm with data collected from a thousand patients.
The data in question concerned six years of each patient, with AI having to 'digest' all the variables that would allow it to be associated with death or heart attack conditions. Once all the data were gathered, the AI ​​was able to predict death situations with an accuracy of 90%, superior to that shown by doctors based on the limited information they have of the patients.
"The algorithm learns progressively with data after several rounds of analysis, and can understand the high-level patterns that must be used to effectively identify who has an event. The result is an individual risk assessment, "Juarez-Orozco said in a statement shared by EurekaAlert.
The researchers' idea is for these AI systems to become useful tools for physicians who can "customize the treatment and ultimately get better results with patients." Miguel Dias/mundophone