The researchers — led by Dr. Luke Oakden-Rayner, a radiologist and PhD student with the Royal Adelaide Hospital and the University of Adelaide School of Public Health — used deep learning to analyze CT scans of 48 patients’ chests. Their findings, published in Scientific Reports, detailed how the deep-learning system went on to predict medical outcomes based on the image patterns.
The algorithm predicted which patients would die within five years with 69 percent accuracy, which is comparable to clinician predictions, according to the researchers. The researchers could not determine which image patterns the algorithm used to create its predictions. However, the most confident predictions were on patients with chronic diseases.
This type of algorithm could help clinicians with early diagnosis and personalized intervention for serious illnesses, according to Dr. Oakden-Rayner.
“Although … only a small sample of patients was used, our research suggests that the computer has learnt to recognise the complex imaging appearances of diseases, something that requires extensive training for human experts,” Dr. Oakden-Rayner said.
More articles on health IT:
House allocates $65M to modernize VA EHRs
Philips opens 1st innovation center in Russia
69% of organizations don’t routinely test security controls, survey finds
