Joining Forces to Develop a New Diagnostic Tool A computer model could help determine how often to test for progression of glaucoma and other chronic conditions Mark Van Oyen, Ph.D., Mariel Lavieri, Ph.D., and Joshua Stein, M.D., M.S.
How frequently should your ophthalmologist ask you the clinic for testing more often than needed,” says Dr. to undergo testing for glaucoma? The answer is imStein. “And patients can become anxious while underportant because patients with glaucoma who are not going testing, which can lead to unreliable results and seen frequently enough are at risk of losing vision that the need to repeat the test.” cannot be restored. Yet too-frequent monitoring can be The computer model, which projects a personalized a burden for patients. “time for next test,” is novel because it is updated each time a test has been performed. The algorithm is A new computer modeling system could help based on data from large clinical trials in combination physicians—especially non-specialists—predict which patients are likely to remain stable and which are likely with patient information—such as eye pressure and to experience worsening disease over a relatively short visual field test results—that accumulates as more tests time. The patient could then be advised when to return are completed. to the clinic for testing and treatment The research team has filed a patent “Our work is based on based on output from an algorithm. for the model, which has applications beyond glaucoma or even eye disease. The project is a collaboration of foundational systems “An algorithm like this can be applied Kellogg glaucoma specialist Joshua to any chronic disease that requires testD. Stein, M.D., M.S., and epideengineering models that ing over time,” says Dr. Stein. “It could miologist David C. Musch, Ph.D., help physicians manage diabetes and M.P.H., and colleagues in Industrial also allowed us to put high blood pressure, or any condition and Operations Engineering at the requiring repeated measurements.” U-M College of Engineering, includa man on the moon.” The research team has validated the ing Mariel S. Lavieri, Ph.D., and Mark P. Van Oyen, Ph.D. model by testing it against data from — Mark Van Oyen, Ph.D. national clinical studies. One such trial, For specialists like Dr. Stein, the Collaborative Initial Glaucoma Treatment Study, led the model provides one more tool to aid with decision by investigators at the Kellogg Eye Center, provided 10 making. “Technology of this sort is not a replacement years of test results on over 600 patients with newly difor the judgment of the physician,” he says. “But taking agnosed glaucoma. “When we compared our algorithm advantage of computer modeling–based innovations can to currently accepted testing practices represented in the absolutely enhance our diagnostic abilities.” trial, it demonstrated better accuracy and decreased de Over 2.2 million Americans have glaucoma, a leadlay in identifying disease progression,” says Dr. Lavieri. ing cause of visual impairment in the United States and worldwide. Because glaucoma often occurs without “Our work is based on foundational systems ensymptoms, a patient can experience irreversible vision gineering models that also allowed us to put a man on loss before being diagnosed with the disease. Any delay the moon,” says Dr. Van Oyen. “This research will lead in treatment can have significant consequences. to more efficient use of resources with more effective “There are also costs to having patients return to patient care.” Innovation in Research
19