Make no mistake: artificial intelligence (AI) exists and is used
in medicine for quite some time. If you do a search on Pubmed,
you will find mention of AI in articles dating back to 1953.
AI refers to software programs that mimic the human body's
cognitive functions, such as reasoning, reasoning and decision-making
making. In recent years, AI has become increasingly mature.
Advanced electronic chips pave the way for superior machine learning
and deep learning. Now it has the real potential to have great
impact on our society as a whole and on medicine in particular.
By carefully analyzing the AI, we discover an escalation in «intelligence»
of these programs. We can have the simplest programs, such as
automated scanners, where patterns in data yield patterns in data
a result (diagnosis). These are useful in automated
blood test programs. The AI software achieves this
by processing large amounts of data and identifying patterns in
this huge amount of data. The patterns may not even be
clear or explainable to researchers.
One of the most common concerns is that AI behaves as a «black
box». This has been reported by Juan Durán and Karin Jongsma in
Journal of Medical Ethics, “Who's afraid of black market algorithms?
box?” in which they discussed issues such as possible bias, the
responsibility, accountability, patient autonomy and the compromised
trust with so-called black box algorithms.
More complex and advanced is AI machine learning where the machine
is trained to navigate through a data set and detect on its own
of the important characteristics that lead to a diagnosis. Numerous
examples have shown clinicians that good predictability with
the educational set does not automatically mean good predictability in
real-life situations, so this type of AI still needs
strict supervision by a human specialist. Finally, there are the deep
neural learning networks that try to closely mimic the complex learning
decision-making processes of which the human brain is capable.
We can say that the current AI programmes are just reaching the
second phase. But the development in AI is very fast and promising
and could even now rebel against the process of making
decisions in ophthalmology. The eye is a particularly good organ for
to implement the care triggered by AI. Most
of our decisions are based on pictures and mathematical formulas,
such as Fourier or Zernike transforms in topographic imaging
of the cornea. Images and mathematics are particularly suitable
for the AI software. We currently have almost all the anatomical
structures of the eye where AI could be applied or is on the threshold
application, both in therapeutic and diagnostic purposes.
AI has applications in retinal diseases such as diabetic retinopathy
retinopathy, retinopathy of prematurity, retinopathy of premature birth, retinopathy of
age-related retinal diseases and age-related retinal diseases
hereditary retinal diseases. In glaucoma, AI can
predict the risk of angle closure and disease progression. Η
AI can also help in cataract care, in ophthalmology, in the
oncology and tele-ophthalmology.
Refractive surgery and corneal laser procedures do not
have been left behind. The introduction of the InnovEyes algorithm in the System
Laser Excimer WaveLight (Alcon) is an interesting tool in the
corneal refractive procedures. Since the introduction of this
AI tool, we can further enhance our clinical trials and
results with LASIK corneal refractive procedures. Now
we're confident that the test group of corneas used
by Alcon and OCULUS for the development of the AI fit tool
closely with patients' corneas in our practice. It also helps
that we can see at the design station, before the surgical
intervention, what the AI tool plans to fix.
AI can be a tool of great value in the diagnosis and
treatment of complex eye problems. Let's embrace AI and
let's work closely with her, but let's remain critical of
to the AI solutions proposed by industry partners. Only
we can measure whether the solutions offered by the
AI tools are productive for our patients. No more
reason to fear the «black box».
