Skip to main content

  • AI, Deep Learning, and the Detection of Systemic Diseases


    Add to My Bookmarks

    For a number of years, investigators have evaluated OCT and color fundus images for their ability to provide retinal biomarkers of cardiovascular and other systemic diseases. Now that artificial intelligence (AI) and deep learning (DL) platforms have entered the arena, where do we stand?

    Tien Y. Wong, MD, PhD, assessed current applications of AI/DL and summed up recent advances in the field of oculomics on Saturday during Retina Subspecialty Day. Broadly speaking, the term “oculomics” refers to the use of measurements obtained from the eye to establish a person’s health status.

    Using AI to estimate risk factors for systemic diseases. In a landmark paper published in 2018, Poplin et al. used AI/DL and color fundus images to predict the risk of developing cardiovascular disease (CVD) over five years.1 Other research groups have been able to use AI/DL to estimate patients’ biological age and evaluate their retinal “age gap”—that is, the gap between patients’ chronological age and their estimated age.

    Using AI instead of existing biomarkers of systemic diseases. Dr. Wong noted that some research groups have been able to use AI/DL with ophthalmic imaging to replace existing biomarkers, such as findings from carotid ultrasound imaging2 or cardiac CT scans3 for CVD and, for dementia, retinal vessel caliber measurements.4

    Using AI to predict specific diseases. Significant progress has been made in using AI/DL to predict CVD and neurological diseases such as Alzheimer and, more recently, Parkinson. For instance, Dr. Wong said, “Multiple groups are using AI/DL on color fundus images and OCT to predict dementia,” he said.5,6

    And with regard to CVD, researchers have been able to detect peripheral arterial disease from color fundus images7 and to predict a patient’s 10-year risk of developing ischemic CDV.8

    Challenges and next steps. But before AI/DL-based platforms can reach their full clinical potential, a number of translational gaps must be addressed and solved, Dr. Wong pointed out.

    In particular, he said, three critical questions remain to be answered: “Who will use the technology—ophthalmologists, optometrists, primary care physicians, or cardiologists? How will clinicians interpret and use the information? And, finally, who will pay for this information?” —Jean Shaw

    1 Poplin R et al. Nat Biomed Eng. 2018;2(3):158-164.
    2 Chang J et al. Am J Ophthalmol. 2020;217:121-130.
    3 Rim TH et al. Lancet Digit Health. 2021;3(5):e306-e316.
    4 Cheung CY et al. Brain Commun. 2022;4(4):fcac212.
    5 Wagner SK et al. BMJ Open. 2022;12(3):e058552.
    6 Cheung CY et al. Lancet Digit Health. 2022;in press.
    7 Mueller S et al. Sci Rep. 2022;12(1):1389.
    8 Ma Y. Sci Bull. 2022;67(1):17-20. doi:10.1016/j.scib.2021.08.016

    Financial disclosures: Dr. Wong. Allergan Singapore: C,L; Allergan: C,L; Bayer Healthcare: C,L,S; Bayer Healthcare Pharmaceuticals: C,L,S; Boehringer-Ingelheim: C; Eden Ophthalmic: C; Genentech: C,L,S; IvericBio: C; Merck: C; Novartis: C,L,S; Oxurion: C; Roche Diagnostics: C,L,S; Samsung Bioepis: C,L; Shanghai Henlius: C; Zhaoke Pharmaceutical: C.

    Disclosure key: C = Consultant/Advisor; E = Employee; EE = Employee, executive role; EO = Owner of company; I = Independent contractor; L = Lecture fees/Speakers bureau; P = Patents/Royalty; PS = Equity/Stock holder, private corporation; S = Grant support; SO = Stock options, public or private corporation; US = Equity/Stock holder, public corporation For definitions of each category, see aao.org/eyenet/disclosures.

    Read more news about Subspecialty Day and AAO 2022.