AI is quite the buzzword these days, yet it holds major promise for medical advancement. While initially we may have thought AI was overhyped, scientists and industry alike are suddenly realizing that Artificial General Intelligence (AGI) is only a few years away. LinkedIn founder Reid Hoffman and NYT columnist Ezra Klein believe we will see AGI within 2-3 years.
The implications are enormous for all medical fields, but in particular for glaucoma, where to-date we have had no way to accurately assess or predict glaucoma progression, much less begin to look at glaucoma pathology. Many patients, despite GAT-based IOP measurements that seem to be within the boundaries of stability, suddenly experience vision loss due to undetected spikes that are occurring regularly outside the office visits.
Dr I. Paul Singh collected and moderated an informative panel at Glaucoma 360 last month that showed the potential for AI across different aspects of glaucoma diagnosis and treatment.
Why We Can't Use AI Yet
But there's a catch. We can't use AI for IOP analysis yet because we don't have the large amounts of continuous IOP data that will give us an accurate interpretation. Sporadic GAT-based measurements cannot flesh out a high-fidelity 24/7 picture of IOP even over the course of years. This is precisely why clinicians have glaucoma patients who seem to be within stable IOP boundaries at check-up but sudddenly suffer vision loss with no warning. There is no indication when regular IOP excursions are happening outside the physician's office.
The lack of 24/7 continuous data means that we cannot even begin to use AI in the same way that endocrinologists use continuous glucose monitoring (CGM) for diabetes data collection. CGM has already led to better prediction of diabetes complications based on the massive amounts of continuous data collected from many different patient segments.
Recently surveys from firms including MarketScope have acknowledged that ophthamologists are calling for exactly this kind of continuous IOP tracking so we can have improved a better understanding of therapy management, including:
compliance insights
the effects of dosage and latency
pharmacokinetics
individualized treatment approaches
The Promise of AI for IOP and Glaucoma Treatment and Research
But for AI -- and especially for machine learning -- we need clean, accurate data. AI algorithms improve with large, high-resolution data sets, and if machine learning begins to build on inaccurate data, it's "garbage in, garbage out."
And while GAT may offer some useful data (this is debatable as GAT machines require periodic calibration, which many clinicians may ignore) these sporadic measurements create gaps that hinder model training and accuracy. AI that usees larger data sets can identify subtle trends and fluctuations that could indicate worsening glaucoma.
At a higher level, we can utilize AI to begin to analyze the correlations between different glaucoma screening methodologies to get a better sense of what corresponding conditions are more likely to lead to vision loss. Imagine being able to determine how OCT and visual field are linked to IOP.
And then there is the research at the highest levels where we can study entire demographic segments with glaucoma to determine how it manifests in that population and likely etiology.
Harnessing AI to increase our understanding of glaucoma is an exciting prospect but we need to follow the proper steps, including having large amounts of accurate data in order to interpret and predict the progression of the disease.
Injectsense: Securing Continuous AI-ready data for IOP Analysis
This is why Injectsense is stepping up with permanent implantable sensors for the general glaucoma patient community. The ultra-miniature sensors, powered by a microbattery, will for the first time offer an accurate IOP Profile, measuring the pressure exerted from the back of the eye near the optic nerve head.
The IOP profile will be based on weeks of frequent and continuous 24/7 measurement where doctors can program sampling rates from each hour to even milliseconds. Opthalmologists can first establish a patient IOP baseline and then measure the effects of glaucoma therapy, from drugs to MIGS or major surgery. With improved data on therapy management, clinicians can then help move the patient back towards a Target IOP Profile.
Comentários