AI will be useful in glaucoma data analysis
Click Here to Manage Email Alerts
Artificial intelligence has many definitions, but as applied to medicine, I like one from Wikipedia. Paraphrased, AI is the use of computers that mimic the cognitive functions of the human mind to enhance learning and problem solving.
The study and application of AI date back to a workshop at Dartmouth University in 1956, so it has been with us for 65 years. As might be expected, the potential of AI was quickly recognized by the Department of Defense, and the DOD invested heavily in AI beginning in 1960.
AI has recently been gaining traction in ophthalmology, primarily as an alternative to the clinician diagnosing and staging of diabetic retinopathy and maculopathy using computer scanning of a patient’s fundus photograph. Evidence is accumulating that AI is competitive with experienced retina specialists in completing this task. The FDA approved the IDx-DR system from Digital Diagnostics for the diagnosis and staging of diabetic retinopathy in 2018.
Another eye care field in which many believe AI can and will play a useful role is the diagnosis and monitoring of glaucoma. Significant clinical skill and clinician time are required in glaucoma diagnosis and management. Careful examination of the optic nerve either directly with ophthalmoscopy or by evaluating stereo photographs, OCT with retinal nerve fiber thickness analysis and visual field outcomes must be interpreted and their individual findings integrated. This requires significant cognitive effort and also significant time, especially when analyzing several years of consecutive testing with variable reliability to look for evidence of progression. In addition, the clinician must integrate other factors including IOP, corneal thickness, blood pressure if abnormal, cerebrospinal pressure if abnormal, other associated comorbidities and family history, to name a few.
Next in line will be genetic testing, which is showing promise in both determining an individual’s risk for developing glaucoma and also how likely a patient with glaucoma will develop progressive nerve damage. AI may also help develop personalized treatment regimens and handicap the risk for progression so that appropriate examination intervals can be determined.
On top of that, technology is being developed that will provide home monitoring of IOP, visual fields, OCT, visual acuity and contrast sensitivity. Some patients may evaluate themselves daily or even multiple times a day at home and then transmit the data to the doctor’s office. All of this data will come cascading into the treating ophthalmologist’s office, and someone or something, such as AI, will need to collate and evaluate this new data source, red flagging the patient data sets that suggest poor glaucoma control and progression of disease so that an office visit can be promptly scheduled.
I believe AI promises to provide clinicians great value in managing their patients with glaucoma. AI will be part of the evidence-based side of the medical care equation. A skilled and experienced clinician will still be needed to collate the evidence and create a personalized treatment plan that takes into consideration each patient’s unique culture, personality and lifestyle needs. In addition, AI will not be able to deliver procedural treatments, whether laser or incisional.
I have no fear that AI will replace me as a clinician and surgeon, and I eagerly anticipate AI computerized programs that can rapidly and accurately plow through the significant data analysis my patients with glaucoma require. For the patient with glaucoma, AI plus a skilled and experienced clinician will be a clear example of one plus one equaling three.
Collapse
Read more about
Click Here to Manage Email Alerts