August 2020


Artificial Intelligence
Artificial intelligence to improve cataract surgery outcomes

by Liz Hillman Editorial Co-Director

Diagram of the architecture of a neural network used to calculate IOL power
Source: Warren Hill, MD


Big data is going to take us to a new level, and I think many recognize that’s where we need to go,” said Kerry Solomon, MD. Is big data and artificial intelligence where it needs to be to aid surgeons in improving cataract surgery outcomes? Not yet, according to Dr. Solomon.
“We don’t have big databases [for cataract surgery],” Dr. Solomon said, noting that he’s talking about datasets with millions of eyes with various diagnostic, preoperative, and postoperative information. “If we were to get large datasets of just a fraction of the 2 million cataracts done per year … we would be able to take our outcomes to a whole new level because there’s so much more we could learn. But right now, those datasets don’t exist.”
As Goh et al. put it in a review article published in the Asia-Pacific Journal of Ophthalmology: “Cataract is one of the leading causes of visual impairment worldwide. However, compared with other major age-related eye diseases, such as diabetic retinopathy, age-related macular degeneration, and glaucoma, AI development in the domain of cataract is still relatively underexplored.”1

What AI could bring to cataract surgery

In general, if datasets were large enough with quality input, Dr. Solomon said they could be analyzed and used to take cataract outcomes “to an entirely different level.”
“Right now, when we are doing cataract outcomes, there are a lot of assumptions that we base our outcomes on. The good news is we’ve gotten so much better with our outcomes today than 10 years ago, but we make an assumption. We make a lot of assumptions based on surgeon factors that are averages across large groups … or perhaps many of us are customizing our own surgeon factors,” he said.
If big data/artificial intelligence is someday employed, Dr. Solomon said cataract surgery could potentially be offered with LASIK-like outcomes and/or better decisions could be made for patients with unique modifiers like post-refractive surgery, for which type of presbyopia-correcting IOL best suits a patient, for best astigmatic treatment.
Warren Hill, MD, cautioned that big data alone won’t improve outcomes, “unless it is also paired with an optimal calculation method.”
“Combined with an appropriate artificial intelligence model, big data allows for the calculation process to be greatly expanded in terms of its breadth and depth,” Dr. Hill said. “For example, many current methods mostly limit possibilities to situations that are already understood, and manual optimization methods significantly limit the number of solution options. Artificial intelligence is well-suited to real world problems for which ideal models are not available. This approach also has enhanced sensitivity for identifying and unraveling complex, non-linear relationships and is free of calculation bias. If based on a high-quality dataset, IOL power selection by artificial intelligence has the potential to take IOL power selection to the next level.”

The status of AI and cataract surgery

When it comes to AI in cataract surgery, there are a few IOL power calculation formulas that draw on neural networks/big data and artificial intelligence. Dr. Hill’s Hill-RBF formula is perhaps the most notable, but there are others, for example the Ladas Super Formula 2.0 and the PEARL-DGS (Prediction Enhanced by Artificial Intelligence and Output Linearization) formula. In addition, the Precision Ladas Universal Super algorithm applies “advanced AI to any static formula” Uday Devgan, MD, explains on his website
The Hill-RBF formula, Dr. Hill explained, was developed with 44 investigators in 22 countries, including mathematicians and engineers at MathWorks in the U.S. He said that 13 different parameters were initially evaluated. Using a genetic algorithm for version 1, they determined the AL, Ks, ACD, and postop spherical equivalent were the most important items of those evaluated.
“As the size of the artificial intelligence model fitting dataset is increased, by the time of version 3 of the Hill-RBF method, we were able to improve outcomes by also taking into account the WTW, LT, CCT, and patient gender,” he said, adding that a beta version of version 3 is now available at It will be incorporated into the LENSTAR EyeSuite software (Haag-Streit Diagnostics) this fall.
In terms of accuracy, Dr. Hill said “an ‘in-bounds’ indication for the Hill-RBF method tells us that there is sufficient data to carry out the calculation at a predicted 90% ±0.50 D accuracy level for a series of patients.” Dr. Hill explained that an “out-of-bounds” indication means data is insufficient to predict the outcomes at this level of accuracy. He said the pairwise boundary models for this part of the calculation were expanded with the number of out-of-bounds indications significantly reduced in version 3 of the Hill-RBF.
Dr. Hill said formulas driven by artificial intelligence are also still reliant on input of quality preop measurements. Stability of the ocular surface for keratometry, for example, is important to ensure accuracy of outcomes.
“An unstable ocular surface is likely to make the Ks unreliable. A 1 D error at the corneal plane is a 1 D error at the spectacle plane. In our practice, we optimize the ocular surface for several weeks prior to biometry,” Dr. Hill said.
While Dr. Hill’s formula uses a large dataset, Dr. Solomon envisions greater potential with larger datasets that have biometry, topography, OCT, tomography, and postop outcomes.
Dr. Solomon said the collection of this data is beginning with programs like Veracity Surgical (Carl Zeiss Meditec). While Veracity has tens of thousands of data points to enhance surgical decision making through its program, Dr. Solomon said when it gets into the millions, there will be greater ability to make better decisions.
“How do we get there? It has to be more widely distributed, more widely accepted, offices have to use it, etc. For that to occur, more offices have to accept it, perhaps the process has to become more streamlined and more intuitive and user friendly,” he said. “Do I think we’ll get there with Veracity? Absolutely. Will there be other products beside Veracity? Absolutely.”

What’s needed for AI to benefit cataract surgery

Further refinements are needed to best apply AI for IOL power calculations, Dr. Hill said, explaining that the main limitation is the accuracy of metrics used for model fitting. He said preop keratometry and the postop refraction are two areas with the greatest variability. He added that as technology used to collect these and other measurements becomes more accurate, artificial intelligence models will offer more predictable outcomes.
Dr. Solomon said for true potential to be realized datasets also need to grow.
“We have to get rid of the data that’s in silos. Every office has data, whether in a paper chart, an EMR chart, sitting in your biometer, topographer, or OCT machine, we all have data on our own patients. Most of us don’t have access to our own data and most of us don’t have access to other people’s data, and the only way we’re truly going to have big data is if we can break down these silos and allow us to put this data in a cloud so we can share this information so we can all benefit from it,” Dr. Solomon said.
Dr. Solomon also cautioned against the notion that AI is a “computer telling me what to do.”
“Nothing could be further from the truth,” Dr. Solomon said. “All the computer is doing is helping to assimilate a bunch of information and taking the doctor’s logic and the doctor’s own preferences and helping them assimilate it to simplify the decision making they would ultimately arrive at, if they had the right information. But how would an individual practitioner be able to assimilate information from a million cases? You wouldn’t.” That’s where AI comes in.

At a glance

• Artificial intelligence (AI) is making inroads in cataract surgery to improve outcomes, namely through IOL choice and power calculations.
• There are several IOL power calculation formulas that are based on analysis of large datasets analyzed by AI.
• To realize the true potential of AI in cataract surgery, experts think much larger datasets are needed.

About the doctors

Warren Hill, MD

East Valley Ophthalmology
Mesa, Arizona

Kerry Solomon, MD
Carolina Eyecare
Mount Pleasant, South Carolina


1. Goh J, et al. Artificial intelligence for cataract detection and management. Asia Pac J Ophthalmol. 2020;9:88–95.

Relevant disclosures

: Haag-Streit Diagnostics
Solomon: Carl Zeiss Meditec



Artificial intelligence to improve cataract surgery outcomes Artificial intelligence to improve cataract surgery outcomes
Ophthalmology News - EyeWorld Magazine
283 110
220 183
True, 8