Boon or Bane Debate of Artificial Intelligence in Ophthalmology Teaching and Clinics

Incorporating machines capabilities to mimic the cognitive abilities of humans, including reasoning, learning and problem-tackling, is largely what artificial intelligence (AI) comprises of AI in its weak format has been existing in tasks such as facial recognition or voice commands in our day-to-day living for long now. More compelling and enhanced presence of AI that is capable of human-level intelligence is now treading into every aspect of professional, industrial, financial, and various other sectors. Machine learning, which enables machines to learn from very large data sets and its subset, Deep learning, which involves neural networking algorithms to replicate human cortical functioning seem to be in the forefront in most of the applications. These are here to stay, with the ease afforded by AI use simplifying complex tasks by automating them and yielding results in a significant low framework of timeline. However, the intelligence of the decision-making needs human expertise to ponder over the accuracy of the outcome delivered. If this concept can be recognized in all its merits, incorporation of AI into every aspect of our functionality would make it more dependable and trustworthy. This will help to correlate with the debate whether AI is a boon or bane to our existence in times we are living in. AI has a multifaceted role in ophthalmic education and training and in ophthalmic healthcare.1,2 Key aspects of AI in ophthalmology teaching include teaching and training enhancement modules involving virtual reality and augmented reality, which are largely employed in surgical simulations for structured surgical teaching/learning to master better microsurgical skills. This also serves to provide an objective method of providing feedback on the trainee performance, help simulate various surgical scenarios and thereby facilitate adaptive learning in a clinical resource-restricted environment. AI can generate clinical case scenarios that can help trainees learn diagnosis and decision-making skills in their postgraduate/fellowship training. Machine learning has put into the forefront AI tools that can assist in or interpret and grade clinical severity in conditions such as diabetic retinopathy, glaucomatous disc changes, corneal ectasia detection and progress monitoring, etc. Thus, given the rapid developments in the evolution of AI and its penetration into the ophthalmology teaching, training and diagnostic imaging, these intelligent teaching systems, virtual case study modules, and fundus interpretation modules are only going to get more robust with time. AI in ophthalmic practice can help potentiate patient communication, increase workflow and clinical decision-making. Incorporation of AI algorithms is being widely explored in several ophthalmic subspeciality fields, including diabetic retinopathy, age-related macular degeneration, retinopathy of prematurity, retinal vascular occlusions, ocular oncology, glaucoma, neuro-ophthalmology, keratoconus, cataract, refractive errors, retinal detachment, and strabismus. AI enhanced analytics for monitoring intraocular lens (IOP), predicting treatment responses to antiglaucoma medications, optic disc analysis for glaucoma diagnosis and progression, optical coherence tomography imaging of optic disc, angle assessment, retinal nerve fiber layer and ganglion cell layer analysis, visual field analysis to identify suspicious fields, differentiate glaucomatous from nonglaucomatous visual fields are some of the ones available for glaucoma care facilitation. Advanced AI algorithms also help to predict futuristic visual field changes long term, along with a personalised inputs which can help to customise target IOP as per the patient’s disease severity. Several AI platforms that investigate the detection of diabetic retinopathy changes and progression, using various AI applications, have set the stage for futuristic fundus screening and other related retinal evaluation diagnostics for diabetic patients. AI applications have also facilitated the development of quantitative scores for defining features of retinopathy of prematurity, vascular occlusions, macular pathologies, biometry applications, cataract density detection and grading, ocular surface tumour diagnostics with image characterisation, strabismus surgery planning, phakic IOL vault predictions, microbial keratitis imaging and corneal ectasia detection and evaluations. While AI is set to revolutionise ophthalmic health care approach, the concerns in encroachment on data privacy, accuracy of the AI predictions, and the resultant diminishing of human intelligence with increasing dependency on machine algorithms might proliferate to worrisome aspects in future. AI in education facilitates a paradigm shift from passive learning to interactive, AI-driven learning, which enables fellows and trainees to master diagnostic, decision-making, treatment planning, and surgical training in a risk-free, simulated environment. Nevertheless, it is imperative to realise that ophthalmology practice should stay centred around the decisive aspects of the human mind applications rather than artificially driven ophthalmic health care. AI can thus be a boon and a bane, and it is up to us as the ophthalmic fraternity to determine the intensity of its impact in our daily practice.