AI Model Can Predict Long-Term Mortality From Chest Radiographs

A convolutional neural network (CNN) can predict long-term mortality from chest radiographs (CXRs), according to a study published online in JAMA Network Open.

Michael T. Lu, M.D., M.P.H., from Massachusetts General Hospital in Boston, and colleagues developed and tested a CNN, named CXR-risk, to predict long-term mortality from CXRs. Data were included from the screening radiography arms of the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO; development, 41,856 patients; testing, 10,464 patients). External testing was performed using data from the screening radiography arm of the National Lung Screening Trial (NLST; 5,493 participants).

The researchers identified a graded association between the CXR-risk score and mortality among participants from the PLCO testing cohort and the NLST testing cohort. The very high-risk group had mortality of 53.0 and 33.9 percent in the PLCO and NLST cohorts, respectively, which was higher than the very low-risk group (unadjusted hazard ratio, 18.3 for PLCO and 15.2 for NLST). The correlation persisted after adjustment for radiologists' findings and risk factors (adjusted hazard ratios, 4.8 and 7.0 for PLCO and NLST, respectively). The results were comparable for lung cancer death, noncancer cardiovascular death, and respiratory death.

"To our knowledge, this was the first report of deep learning to predict long-term prognosis from chest radiographs," the authors write. "Further research is necessary to determine how this can improve individual and population health."

Several authors disclosed financial ties to the pharmaceutical and medical technology industries.

-----------------------------------------------------------------------
Editor's Note:

If you have any suggestion to the content, 

please email: Julia.Zhang@ubmsinoexpo.com