AI adoption (what, how)
AI is so ubiquitous these days that it is almost impossible to do anything without the assistance of an AI machine. What’s even more surprising is that we are not even aware how present AI is in our lives. From the videos we watch on YouTube and TikTok, to the movies we watch on Netflix. The cars that self-drive and the robot vacuums that clean our house. Thanks to the ground-breaking steps taken by B&VIIT Eye Center, AI is now also a part of the vision correction industry.
Unlike humans, machines do not fatigue, get tired or sleepy, and are not affected by time zones or emotional states. Since pre-operation examinations require very precise measurements that need to be verified and re-verified, it would stand to reason that we could trust a machine to produce reliable results.
During consultation and testing, optometrists and ophthalmologists would constantly have to check the status of the patient while evaluating data and asking themselves questions such as: “Is refractive surgery possible for this patient?” “What surgical option is appropriate?” and “What will the vision level be after the surgery?” Now, thanks to the incorporation of AI in ophthalmology, they do not have to ask these questions anymore. The AI system now takes care of that, answering the questions more accurately and at much quicker speed.
Ophthalmological evaluations involve highly complex images and figures and deciphering them is an extremely difficult process, especially considering the 3D nature of some of the images. It takes an experienced ophthalmologist decades of training to be able to recognize irregular corneas, which would potentially disqualify a patient from receiving surgery. After a large amount of data is used to train an AI machine, it would be able to recognize and evaluate the same potential irregular corneas, without requiring many, many years of training. This will essentially allow ophthalmologists to not only make faster decisions, but will also allow for more accuracy regarding evaluation. This is great for doctors and patients. It narrows the proficiency gap for doctors who have less experience and it provides patients with peace of mind, knowing their vision correction procedure choice is safe and personalized.
There is often quite a lot of confusion on exactly how vision correction works. There is an assumption that you can just say: “I want LASEK.” Or “I think SMILE is the one I want.” Unfortunately, in reality the vision correction process does not work like this.
There is a two-hour long pre-surgery and evaluation stage where the clinic will evaluate various aspects of the eye and its condition. The examination phase is the most important phase, because it is during this time that the ophthalmologists determine which vision correction procedure each individual may be suited for, as there is a plethora of procedures available: SMILE, LASIK, LASEK, ICL, Xtra SMILE, XTRA LASIK etc.
B&VIIT operates at a high standard of safety, and is thus sometimes conservative with their recommendations. Sometimes, a patient may have their heart set on one procedure, such as SMILE for example, but the clinic may deem it too risky, and therefore recommend another procedure. An example of this is the conservative standard for residual cornea thickness maintained by B&VIIT, which would preclude certain individuals from LASEK. So, what does this have to do with AI?
According to the report ‘Artificial intelligence and life in 2030’ published in Nature Medicine, by Jeffery de Fauw. Joseph R. Ledsam, Artificial intelligence is the activity devoted to making machines intelligent, and intelligence is that quality that enables an entity to fiction appropriately and with foresight in its environment.” With this we can accept that AI in medicine and more specifically ophthalmology, is currently focused on primarily assisting doctors and not on replacing doctors. The goal is to make the vision correction procedure so standard and run-of-the-mill that it will become almost as synonymous as going to get a filling at the dentist. With our current AI this is what we are moving ever close to, day by day.