Researchers at the University of Nottingham have developed artificial intelligence that can analyse patients’ health status and predict their eventual death with extremely high accuracy.
Researchers at the University of Nottingham have developed artificial intelligence that can analyse patients’ health status and predict their eventual death with extremely high accuracy. To do this, the university team “trained” the AI with data from over half a million English patients belonging to the period from 2006 to 2016. During this decade 14,500 study participants died mainly of cancer, heart attack or diseases lungs, elements that artificial intelligence was able to predict with an accuracy of 76 percent.
“Preventive care is becoming a priority in the fight against deadly diseases, so we have worked for years to improve the accuracy of computerised predictions applied to the population,” said Stephen Weng, head of the research team and the study. “Most applications are based on a single disease, but predicting death due to various diseases is extremely complex.” Artificial intelligence programmed by university researchers, on the other hand, is able to analyse patient data and restore a probability of death due to various factors, mainly cancer, heart problems or lung disease.
According to Weng, the research represents an important step forward in predicting premature deaths, for which factors such as age, gender, smoking and family cancer diagnosis may play a role in the probability of death. But the elements that potentially affect this probability are innumerable and can also include lifestyle, body fat, diet and physical activity, all data that only an algorithm is able to analyse on a large scale to provide data useful for doctors and patients to avoid premature problems.