Interview: "Clinicians are inconsistent—AI is not"

Search Dental Tribune

Interview: “Even the most experienced clinician is inconsistent—AI is not”

Dr Kyle Stanley maintains a private practice in Beverly Hills in the US, where he focuses on implant surgery and prosthetics. (Image: Kyle Stanley)

Dr Kyle Stanley is a specialist in implantology, founder of a company transforming patient care through artificial intelligence (AI) and a passionate advocate for mental health in the dental profession. On 16 September, he will lecture on the future of dentistry powered by AI at the Osstell ISQ Online Symposium, which attracted 2,000 attendees last year. In this interview, the expert discusses how AI supports dental professionals in different areas such as radiology, practice management, smile design, the laboratory and dental insurance. He also talks about how AI is going to help individualise treatment plans in the future and how it can benefit the mental well-being of dentists.

Dr Stanley, in your Osstell ISQ Symposium lecture, you will speak about different applications of AI in dentistry. Could you please explain how AI is enhancing different aspects of dental care and what the benefits for dental professionals and their patients are?
Patient care begins with diagnosis, and in dentistry, diagnosis starts with radiographs. Computer vision AI has proved a potent dental diagnostic tooland diagnostic consistency is the source of much of AI’s potency. That’s because dentists have an unfortunately large consistency deficit. Many will be familiar with a 1997 Reader’s Digest story written by an investigative journalist who visited 50 different dentists across the US over four months, showed them each the same radiographs of his teeth, and received 50 different diagnoses and treatment plans. That experience has been validated in numerous studies over the years, including one recently by the Dental AI Council, in which 136 dentists reviewing the same full-mouth radiographs never delivered better than 50% diagnostic consensus on any given tooth. They also produced widely divergent full-mouth treatment plans, ranging in cost from US$300 to US$36,000.

These inconsistencies can be attributed to factors such as differences in training, experience and state of mindbut whatever the cause, it’s quite clear that even the most experienced clinician is inconsistentAI is not. Once trained, computer vision systems for diagnostics will perform the same job the same way every time. They are consistent above all else. Because a computer vision diagnostic system’s intelligence is so narrowly focused, its performance is unaffected by the preconceptions, bias or fatigue that affect even the most expert human radiologist. This makes them ideally suited to providing practitioners with chairside diagnostic second opinionsvalidation that delivers cascading benefits that have an impact on everything, including treatment planning, patient trust, insurance claim approvals and medical liability.

Of course, these systems are also very fast. In just a few seconds, a computer vision system can deliver diagnoses on thousands of radiographs. Although that kind of speed may not be particularly necessary in a one-on-one patient consultation, it becomes very useful in tasks like evaluating every historical radiograph in an office’s practice management system to pinpoint the provider’s diagnostic strengths and weakness, surfacing missed diagnoses in a patient population and, over time, establishing a clear, elevated standard of care. These sorts of systems are on the verge of entering patient-facing clinical environments, but are presently pending US Food and Drug Administration (FDA) approval.

Since practice management is not a patient-facing area, however, FDA approval is not required, and diagnostic radiology AI systems are already delivering noticeable benefits in that arena. Computer vision diagnostic systems are converging with practice management systems, where historical radiographs deliver a wealth of actionable insights into both patient health and practitioner performance. This application is useful in solo practices and especially in larger groups and dental service organisations, where the high volume of data being generated makes it particularly useful.

“I stand behind AI because I am always looking for ways to improve the quality of the work I do”

The same diagnostic AI technology is being employed by insurance carriers to review radiographic claims evidence. Historically, carriers have always received far too many claims with radiographs to be able to actually validate them all. Applied in utilisation management, computer vision made it much easier to diagnostically validate every claim. These systems review all radiographic attachments and compare their diagnostic findings with diagnostic codes listed in the claim. If everything matches up, the claim is paid. If the AI system’s findings and codes don’t align, then the claim is passed to an expert human reviewer. This means that investment in human capital goes only towards claims actually worthy of review, vastly reducing the chances that a bogus claim will be approved.

Of course, as a practitioner with an interest in aesthetic dentistry, one of the areas of AI development in dentistry that excites me most is smile design. Systems employing generative neural networks, like those used to create deepfakes and Snapchat filters, are entering deployment, not only allowing doctors to automatically create a realistic rendering of an ideal smile that they can show to patients, but also translating the approved rendering into practical restorative prosthetic designs for milling within the CAD/CAM workflow.

To be honest, in the examples and explanations I have outlined, I have only just scratched the surface in addressing the ways that AI is going to have an impact on the dental workflow. And as AI continues to prove its value, there will be increasing pressure on every kind of enterprise operating within the web of dental business operations to adopt new methods and processes specifically tailored to serve AI systems and maximise the production of data from which the AI can draw insights to drive efficiency, consistency and quality across every aspect of dentistry.

The COVID-19 pandemic has made digital solutions even more important for dentistry. Have you noticed an increase of AI usage in the dental community since the onset of the pandemic?
Certainly. With many dental offices closed, many of the practice owners I saw went from working in their practices to working on themlooking for efficiencies and developing a more open mind towards technologies, including AI, as well as towards teledentistry.

“It’s not doctor versus AI”

What would you tell dental professionals who are still hesitant to jump on the AI bandwagon?
Whether we want to accept it or not, AI is already hereand it is absolutely certain to play a critical role in the future of dentistry. Those people who don’t want to jump on the bandwagon tend to be people who view it as a threat or who think they are infallible or perfect at their jobs. The “infallible” dentists cannot be swayed, but to those who view it as a threat, I would start first by saying that I come from a family of dentists, and I would never stand behind anything that would jeopardise our livelihood. I stand behind AI because I am always looking for ways to improve the quality of the work I doand AI will allow me to do that. It is not taking away my livelihood. If anything, it is making me a better dentistsomeone my patients should put more trust in, because the work I do is being assisted by a device that prevents me from making mistakes that every human is subject to making.

And that’s really the key: AI is an assistive tool. It’s not doctor versus AI. It is doctor together with AI. By working together, we get better. I like to compare it to intelligent cruise control in modern cars: it will help keep you in the lane, it may tap your brakes if you’re distracted and approaching a stopped car too fast, but you hold the steering wheel and you have the final say when it comes to where you’re going and what route you take to get there. And along the way, AI will help limit your liability, build your patients’ trust, and help you do more and better dentistry.

Together with two partners, you have founded your own company—named Pearl—which specialises in AI solutions for dental professionals. What is your role within the company, and how do your products support dental professionals in their everyday practice?
At Pearl, I am the chief clinical officer. What this means is that I am the dental brain on the team. When products are being created, I make sure they are clinically sound and make sense for the business of dentistry. I also manage a team of doctors who help annotate images that help train AI models. In addition, I represent the company when interfacing with members of the dental community, speaking to doctors, group practice owners and office staff on podcasts, at conferences and in person. Pearl has an array of products which serve different industry constituents: Practice Intelligence is our clinical performance management solution, Claims Review and Pearl Protect are our solutions for the insurance market, Smart Margin and Prep Assess serve practices and dental laboratories with efficiencies within the restorative fabrication workflow, and Second Opinion is our patient-facing radiological assistive software, which is set to launch in the EU as we await FDA approval in the US. We also have an educational tool which is beginning to be deployed at various dental schools.

“We can expect AI to touch every aspect of dentistry”

Even though AI is already widely used in dentistry and many other aspects of our daily life, surely we have not exploited its full potential yet. In your opinion, what can dental professionals look forward to in the years to come?
We can expect AI to touch every aspect of dentistry. In the future, all diagnoses and treatment planning will be assisted with AI, but more importantly our treatment plans will become increasingly individualised. What this means is that two people with the same diagnosis could get different treatment plans based on their genetics, musculature, bacteria flora, bone density, tooth shape and any other input that we can measure. We can also expect to see AI helping to make dentistry more predictive and, therefore, more preventive.

Is there anything else you would like our readership to know?
Those that know me from social media (@drkylestanley) know that I am a huge proponent of fostering mental health within dentistry. People rarely mention it, but I believe AI will have a real benefit in that area, because it will decrease liability, increase consistency of diagnosis and treatment planning, and tackle many of the mundane tasks that contribute to doctor decision fatigue in our industry. It is a pleasure for me to be able to help introduce new technology that may help reduce dentist burn-out and create a better future for this great profession.

Editorial note: Dr Kyle Stanley’s symposium session, titled “The future of dentistry powered by AI”, will be broadcast on 16 September at 10 p.m. CEST. Participants will be able to earn a continuing education credit by answering a questionnaire after the lecture. Dental professionals who would like to attend the presentation may register on the Osstell Campus.

AI Artificial intelligence Dental implant Dentistry Dr Kyle Stanley ISQ Online Symposium Osstell

Leave a Reply

Your email address will not be published. Required fields are marked *