A group of Australian researchers has developed a new artificial intelligence (AI) system, which can predict whether or not a patient will die soon by simply looking at images of the person's organs. The research paper, which was published recently in the Nature journal Scientific Reports, is expected to help physicians with early diagnosis of serious illness.
The AI system was built by researchers at the University of Adelaide, who used artificial intelligence to analyse medical images from the chests of 48 patients. According to the researchers, the computer-based analysis predicted which patients would die within five years with 69 percent accuracy.
"The accurate assessment of biological age and the prediction of a patient's longevity has so far been limited by doctors' inability to look inside the body and measure the health of each organ," Dr Luke Oakden-Rayner, the lead author of the research paper, said in a statement. "Our research has investigated the use of 'deep learning', a technique where computer systems can learn how to understand and analyse images."
While the researchers couldn't identify exactly what in the images helped the computer system make its predictions, it found the most success with patients suffering from severe chronic diseases such as emphysema and congestive heart failure.
According to the researchers, the AI system is capable of predicting medical outcomes by incorporating large volumes of data and detecting subtle patterns, something that doctors are not trained to do.
"Our research opens new avenues for the application of artificial intelligence technology in medical image analysis, and could offer new hope for the early detection of serious illness, requiring specific medical interventions," Oakden-Rayner said.
The researchers hope that they can use the same technology to predict other major medical conditions, such as the onset of a heart attack.
In a similar recent development, a team of Google Scientists claimed that the AI technology the company uses for its self-driving car project, can detect cancer faster and more accurately than human doctors.
The researchers claimed that their framework could "automatically detect and localise tumours as small as 100 ×100 pixels in gigapixel microscopy images sized 100,000×100,000 pixels."
