Opportunistic detection of type 2 diabetes using deep learning from frontal chest radiographs.

Journal: Nature communications
Published Date:

Abstract

Deep learning (DL) models can harness electronic health records (EHRs) to predict diseases and extract radiologic findings for diagnosis. With ambulatory chest radiographs (CXRs) frequently ordered, we investigated detecting type 2 diabetes (T2D) by combining radiographic and EHR data using a DL model. Our model, developed from 271,065 CXRs and 160,244 patients, was tested on a prospective dataset of 9,943 CXRs. Here we show the model effectively detected T2D with a ROC AUC of 0.84 and a 16% prevalence. The algorithm flagged 1,381 cases (14%) as suspicious for T2D. External validation at a distinct institution yielded a ROC AUC of 0.77, with 5% of patients subsequently diagnosed with T2D. Explainable AI techniques revealed correlations between specific adiposity measures and high predictivity, suggesting CXRs' potential for enhanced T2D screening.

Authors

  • Ayis Pyrros
    DuPage Medical Group, Radiology. Electronic address: ayis@ayis.org.
  • Stephen M Borstelmann
    University of Central Florida School of Medicine, UCF College of Medicine, 6850 Lake Nona Blvd, Orlando, FL 32827. Electronic address: sborstelmannmd@gmail.com.
  • Ramana Mantravadi
    Brainnet, Inc., West Harrison, NY, USA.
  • Zachary Zaiman
    Department of Computer Science, Emory University, Atlanta, GA, USA.
  • Kaesha Thomas
    Department of Radiology, Emory University, Atlanta, GA, USA.
  • Brandon Price
    Department of Radiology, Florida State University, Tallahassee, FL, USA.
  • Eugene Greenstein
    Department of Cardiology, Duly Health and Care, Downers Grove, IL, USA.
  • Nasir Siddiqui
    DuPage Medical Group, Radiology.
  • Melinda Willis
    DuPage Medical Group, Radiology.
  • Ihar Shulhan
    EPAM, Inc, Newtown, PA, USA.
  • John Hines-Shah
    Duly Health and Care, Department of Radiology, Downers Grove, IL, USA.
  • Jeanne M Horowitz
    Department of Radiology, Northwestern Memorial Hospital, Northwestern University, Chicago, Illinois.
  • Paul Nikolaidis
    Northwestern Memorial Hospital, Northwestern University, Radiology.
  • Matthew P Lungren
  • Jorge Mario Rodríguez-Fernández
    University of Illinois at Chicago, Department of Neurology.
  • Judy Wawira Gichoya
    Department of Interventional Radiology, Oregon Health & Science University, Portland, Oregon; Department of Radiology and Imaging Sciences, Emory University, Atlanta, Georgia.
  • Sanmi Koyejo
    Department of Computer Science, University of Illinois at Urbana-Champaign, Champaign, Illinois.
  • Adam E Flanders
  • Nishith Khandwala
    Stanford University.
  • Amit Gupta
    Department of Cardiology, SKIMS, Srinagar, India. Electronic address: amitcardio12@gmail.com.
  • John W Garrett
    From the Departments of Medical Physics (R.Z., X.T., C.Z., D.G., J.W.G., K.L., S.B.R., G.H.C.) and Radiology (M.L.S., J.W.G., K.L., S.B.R., G.H.C.), University of Wisconsin-Madison School of Medicine and Public Health, 1111 Highland Ave, Madison, WI 53705; and Department of Radiology, Henry Ford Health System, Detroit, Mich (Z.Q., N.B.B., T.K.S., J.D.N,).
  • Joseph Paul Cohen
    Unity Health Toronto (Verma, Murray, Straus, Pou-Prom, Mamdani); Li Ka Shing Knowledge Institute of St. Michael's Hospital (Verma, Straus, Pou-Prom, Mamdani); Department of Medicine (Verma, Shojania, Straus, Mamdani) and Institute of Health Policy, Management, and Evaluation (Verma, Mamdani) and Department of Statistics (Murray), University of Toronto, Toronto, Ont.; University of Alberta (Greiner); Alberta Machine Intelligence Institute (Greiner), Edmonton, Alta.; Montreal Institute for Learning Algorithms (Cohen), Montréal, Que.; Centre for Quality Improvement and Patient Safety (Shojania), University of Toronto; Sunnybrook Health Sciences Centre (Shojania); Vector Institute (Ghassemi, Mamdani) and Department of Computer Science (Ghassemi); Leslie Dan Faculty of Pharmacy (Mamdani), University of Toronto, Toronto, Ont.; Department of Radiology, Stanford University (Cohen), Stanford, Calif.
  • Brian T Layden
    Department of Medicine, Division of Endocrinology, Diabetes, and Metabolism, University of Illinois at Chicago, Chicago, Illinois, United States of America.
  • Perry J Pickhardt
    University of Wisconsin Medical School, Department of Radiology, Madison, Wisconsin, United States.
  • William Galanter
    Department of Medicine, College of Medicine, University of Illinois at Chicago, Chicago, Illinois, United States of America.