As a practicing ophthalmologist, I understand our field’s innate tendency to be drawn towards new shiny gadgets and gizmos like a moth to light. However, this inclination often results in less attention being paid towards technologies that are more abstract. It creates an odd juxtaposition where we may be offering the highest quality medical and surgical solutions against a backdrop of outdated software solutions and workflows.
The tides are changing though, and I am excited for what I believe will be some fundamental changes in the approach to modern eye care. So why the tide change? There are many factors at play, including decreasing reimbursements, increasing overhead, increased staff turnover and higher patient expectations for a smooth experience from appointment scheduling to outcomes. This is all driving us toward the “new Big E” as Alcon’s former CMO Dr. Stephen Lane put it on The History of Eyecare podcast last year. No longer is emmetropia enough. It’s efficiency that will allow practices to survive and thrive in this harsh new world.
Enter AI, the often conflated, overhyped, panacea-pedaled technology understood by few. Despite the hype, there are real use cases already in practice, such as ambient listening AI scribes, fundus/OCT image analysis, IOL formulae/design and so-on.
The Critical Role of Medical Intake
I have come to believe that medical intake is the cornerstone of high-quality patient care. No different than a botched corneal paracentesis and main incision making the rest of a cataract surgery miserable, a botched medical intake creates more friction, more work and higher risk for errors downstream.
After speaking with various leaders in the field, I was surprised to find we still have a ways to go before we can say we’ve made pre-engagement efficient. The solutions that have emerged across EHR vendors and third party software vendors in ophthalmology look something like this:
- Patients can create an account, log into a patient portal and upload insurance information, payment, ID, demographics and basic medical history like allergies, medications, medical and surgical history. AI “agents” can be used to engage patients autonomously to gather some patient information and help with patient appointment scheduling.
I do think it is important to note that outside of software solutions, there are the much more resource-intensive call centers. These may be in-house or outsourced through business process outsourcing (BPOs), and they manually call and/or text patients to assist with more administrative tasks and clinical triage.
Leveraging the Pre-Engagement Space
What I have yet to come across is a software solution that is able to autonomously gather a patient’s medical history and fully leverage the opportunities in the pre-engagement space.
What do I mean by this? Most software solutions are still leaving it to practices to gather the patient’s chief complaint(s) and history of present illness(es). Further, they are not providing patient-specific education or screening for clinical trials. Lastly, they do not screen for social determinants of health (SDOH) and have limited capacity for engaging patients in their native language.
Using AI Tools to Their Advantage
AI has turned into a catch-all term. It’s important to understand that, under this umbrella, there are specific tools that are better suited for certain purposes.
- Machine learning/deep learning: A subset of AI that allows systems to learn from data without being explicitly programmed. Examples of machine learning are Netflix recommending what show to watch next based on what you’ve watched and liked, or image recognition like when Apple Photos can suggest who’s in your photo albums based on images you’ve previously tagged with names.
- Natural language processing (NLP): A branch of AI that enables machines to understand, interpret and generate human language. This is the technology behind things like chatbots, sentiment analysis and machine translation. Examples of NLP are Amazon’s Alexa, Apple’s Siri, and search engines that understand query intent – which is very relevant to medical intake.
- Computer vision: A field of AI that enables computers to “see” and interpret visual information from the world. Think self-driving cars and all of the ophthalmic imaging software analysis solutions out there for things like screening for diabetic retinopathy or ocular coherence tomography analysis.
- Expert systems: AI systems that are designed to solve problems within a specific domain, mimicking the decision-making ability of a human expert. A rather famous medical expert system was MYCIN. It was a rule-based system that used a knowledge base of approximately 600 “if-then” rules, derived from human experts, to provide diagnoses and justify its recommendations. While it demonstrated expert-level performance comparable to human specialists, MYCIN was never used in clinical practice due to ethical concerns and usability issues of the time (like the need to enter data manually via a terminal, which was a barrier for busy clinicians).
There are many more branches of AI and my descriptions are limited, but hopefully these examples bring clarity to the following discussion.
Using the Latest Technology
In eye care, AI applications have largely focused on image analysis (computer vision), which makes sense, as a great deal of our workflow focuses on visualizing structures and analyzing them. The other big one that has been a very hot topic the last few years is the ambient scribe, which can analyze a clinical conversation in real time and then record and summarize it. This is probably one of the most well-known clinical applications of natural language processing. Lastly, there are a number of IOL calculation formulas like Hill-RBF and Kane that leverage machine learning to fine-tune predicted refractive outcomes.
But we can do better—especially when it comes to medical intake. From what I’ve seen at industry events this year, most of the pre-engagement solutions are still largely administrative, and the clinical applications are fairly limited in scope. We are still relying on resource-intensive processes to triage and work up patients, educate patients and screen patients for clinical trials. Technology can help us address these processes more efficiently and accurately with added functionality.
My Vision for the Future of Pre-Engagement and Patient Intake
I advocate for meeting patients where they are. Naturally, some patients will want or need to speak with a live person to help them with things like scheduling an appointment, obtaining medication refills, processing payments, requesting records or taking their histories. Some people just want to have a human touch, and for certain aspects of care, that will remain forever true. But as Bob Dylan famously wrote, “The Times They Are A-Changin.’
These days, we are accustomed to viewing our test results in patient portals, being able to self-schedule or reschedule appointments on a practice website, order medication refills by text, and upload health insurance, ID and completed intake forms online. A growing number of patients prefer digital interfaces to human ones for its convenience and efficiency. Not everyone has the luxury to wait on the phone or call a practice multiple times throughout the day to get ahold of someone. Ageism often leads people to falsely conclude that older generations do not want to engage with these modes of communication, but it is often those very patients who ask me if my own practice has Epic’s “MyChart.”
If we truly care about meeting patients where they are, we must acknowledge and embrace that an increasing percentage of our patients are not only comfortable with, but demand, more of their care to be digital. And as far as I’m concerned, this is a good thing. It presents the opportunity for practices to become leaner, more efficient, deliver better care and create standardized, universal data sets that can be used for clinical trials, population health tools and better communication between practices, industry partners and doctors.
I envision a future where patients own their health data and are empowered to share as much as they are able and willing with health care professionals so that they can receive the best possible care. Before patients set foot in any health care setting, they should have the opportunity to communicate their complete medical history, their reason(s) for presenting and questions or concerns they’re hoping to have addressed during their visit.
With modern technology and thoughtful application of the various AI tools we have at our disposal, we can help patients tell their story in a guided, individually tailored way while screening for clinical trials and educating patients about their disease and treatment options. Furthermore, we can generate standardized data sets that can push the practice of medicine forward despite the many modern day healthcare hurdles we face today.
Author
