AI solutions company Visulytix* has developed an eye care analysis tool that is being rolled out to selected institutions around the world – both for clinical use and to support investigational studies. The decision support tool, Pegasus, facilitates specialist level decision-making support, at scale, to eye care providers: a first for the industry.
The long-anticipated benefits of AI will surely be a game changer for the screening and care of conditions such as glaucoma, age-related macular degeneration and diabetic retinopathy (conditions that, collectively, affect 500 million people worldwide), but how can AI systems be successfully managed?
Whilst such systems are likely to attract much fanfare, they must be evaluated quantitatively, as well as for their ability to work ‘out of the box’ and in diverse clinical environments.
Eye care professionals such as ophthalmologists and optometrists will be required to embrace the technology and understand, at least at a general level, some of the technical and statistical terms that may be used in product documentation and performance. These include such terminology as sensitivity, specificity and accuracy.
Furthermore, it will be important to evaluate the context in which the AI system should be deployed, as well as any contraindications for its use. It would, therefore, be pertinent to assess the applicability of such an AI system for use in the organisations’ specific environment.
During the course of my travels and numerous discussions with stakeholders, several misconceptions have become apparent regarding the abilities of the first generation of AI decision support systems. It is anticipated that AI systems may grow and are assumed to perform a variety of tasks that a healthcare professional currently undertakes; but this is not the case. Instead, they should be thought of as ‘narrow AI’; that is, technology that can perform a very specific ‘narrow’ task with excellent (and in some cases super-human) ability.
Nevertheless, I have come across many eye care professionals who are extremely tech savvy, willing to embrace this exciting new wave of advancement and adapt their practice environment as AI capabilities develop.
Eye care professionals will still need to obtain the same qualifications and learn the anatomy and pathophysiology around the eye. However, I can envisage a mathematical and technological element to eye care training evolving in the coming years, in order to keep pace with the delivery environment.
Similarly, as AI systems become increasingly diffused into clinical practice, the technology will need to be incorporated into staff induction, training and continuous professional development. Industry will have a key role to play in this respect.
In many organisations, key performance indicators (KPIs) are in place to evaluate staff effectiveness. These may include, for example, the number of glasses sold by the sales team and the total resultant value. Such KPIs are often used to focus the staff member on how they may add value to an organisation.
Similarly, an AI system will have model and performance metrics that should be periodically measured, and internal, local benchmarks should be set. Routine audits and post market surveillance will become commonplace.
In addition, in the same way that staff members undergo regular appraisal—with poorer performers being informed of the gap between what they are and should be delivering—a parallel AI system performance should also be appraised. In the event of AI performance being suboptimal, the organisation should decide whether or not to request an updated version of the system and inform the provider.
As I have mentioned previously, organisational agility and an appreciation of some of these points will permit rapid diffusion of such systems. This will ultimately lead to significant competitive advantage in a crowded ophthalmic care industry.
*Pegasus is currently pending CE mark certification, which is anticipated in the summer of 2018.
Mr Sameer Trikha, MBA FRCOphth
MrTrikha is a consultant ophthalmic surgeon at King’s College NHS Foundation Trust, and founder/chief medical officer at Visulytix, London, UK.