Have you ever gone to the dentist and been unsure if that spot on your tooth the doctor is looking at is really a cavity? Or maybe you’ve gone to get a second opinion, only to have the new practice tell you that you need a crown on a completely different tooth?
Unfortunately, this story is all too common in dentistry — in fact, there’s a well-known story about a Reader Digest reporter who went to see 50 different dentists and received nearly 50 different diagnoses.
That makes dentistry ripe for technological innovation aimed at increasing confidence and accuracy in diagnoses. For many reasons, dentistry is the ideal frontier for AI: Not only does the field produce an abundance of x-rays, but they’re also easy to anonymize and are a great data set for AI/machine learning to scan and learn from. Additionally, the dental field doesn’t have trained radiographers the same way the healthcare industry does, which could make the extra set of “AI eyes” a welcome addition for well-intentioned practitioners.
Los Angeles-based Ophir Tanz, CEO of Pearl, is one such developer hoping dental AI technology can take some of the guesswork out of dentistry, giving both patients and providers peace of mind. The son of a dentist himself, Ophir recognized the potential for AI in the industry, and after successfully standing up contextual intelligence AI company GumGum (now valued at $700M), he’s using the same tech to transform the dental industry.
I connected with Tanz about the future of dentistry and the impact AI could have on patient outcomes and the industry at large.
GN: Why is dentistry the ideal frontier for AI?
Ophir Tanz: The dental field is ripe for AI innovation for a couple of reasons. First, the abundance of radiographic images — patients receive dental x-rays every two years, so there are more dental radiographs in the world than any other form of medical imagery. This is extremely helpful when it comes to developing AI radiologic systems for dentistry because those systems need to be trained on large numbers of radiographs. Second, dentistry has a more entrepreneurial character than other forms of medicine. Most dentists are invested to one degree or another in a practice, so they’re not just doctors but also business owners. A dentist’s primary concern is delivering optimal patient care, which AI helps them do — but it also helps them address the business operations concerns they face as practice owners. The same AI insights that elevate the standard of care and patient outcomes can also be applied to help them make smarter decisions around budgeting, staffing, materials, equipment resourcing, etc. Innovation requires adoption, and dentists are natural early AI adopters because its benefits touch every facet of their work — and because, unlike the majority of doctors in other fields, dentists are business owners, so they have both the authority and impetus to invest in AI.
GN: There are similar applications rolling out in other medical spheres. Can you give us an overview of how AI is being used to read scans across the medical ecosystem?
Ophir Tanz: There is a wide range of AI technologies being applied in other areas of medicine — not only in radiologic applications but in intake, triage, biologic testing-based diagnostics, predictive diagnostics, etc. Talking specifically about AI-based analysis of medical imagery, thousands of radiologic AI systems that have been developed over the past 15 years. The vast majority of these systems have come out of research institutions. Not all of these systems have proved useful; many of those that could be useful are effectively redundant (i.e. they perform the same task with more or less the same outcome), and not all of those where both effective and novel have found their way past the regulatory and commercial hurdles to application in the real-world. There are currently around 350 FDA-approved medical devices that apply AI in some capacity, and the vast majority of these perform some degree of analysis of medical imagery. Most help automate repetitive tasks, like anatomical segmentation. However, there are plenty of AI-powered imaging systems that perform diagnostic functions. Whatever their use — oncology, neurology, cardiology, ophthalmology, etc. — these devices perform highly specific functions, like detecting a specific condition in a specific part of the body that can be found in a specific type of medical image. As such, the chance that anyone has ever encountered an AI system in the course of their medical care is extremely low. Naturally, this will change as AI technology becomes more generalizable and powerful — but the first medical AI that people, at scale, will ever experience is almost certain to be in a dental office. That’s true not only because people visit the dentist more frequently than they do any other kind of doctor but because we’ve been able to develop systems with broad utility in detecting a comprehensive array of dental conditions.
GN: How is your technology being received by dentists, who may be accustomed to doing things a certain way?
Ophir Tanz: The response we’ve seen from dentists using our solutions has been overwhelmingly positive, but that’s to be expected because early adopters more likely have a more favorable attitude about AI. There are certainly dentists out there who are skeptical. Overcoming that skepticism will require education. Once these skeptics get their hands on the technology and learn more about what it can and cannot do, they’ll realize that AI is not a threat to their profession — that it’s simply a powerful tool that enables them to perform their jobs at a higher level. I expect adoption to accelerate rapidly as AI literacy in dentistry expands and people become more comfortable with the concept of AI diagnostics in general.
This is already starting to happen. We’re selling our real-time radiologic aid, Second Opinion, in Europe, Australia, New Zealand, Canada and various other territories and our AI clinical management solution, Practice Intelligence, is in use in thousands of practices domestically and abroad. These are really transformative solutions, and I believe that as we continue to gain regulatory approval in different parts of the world, dentists will be ready for AI and be quick to incorporate the technology into their daily routines.
GN: How are patients responding to technology rollouts like this one?
Ophir Tanz: Patient response is one of the things that dentists tell us they love most about the technology. Naturally, there’s a wow factor that this technology even exists, and patients appreciate that their dentist is applying the state-of-the-art in delivering care. Then there’s the impact of AI on the patient’s ability to understand their doctor’s diagnosis. Rather than pointing at an indistinct blotch on the radiograph and saying, “It’s hard to make out, but you have a cavity here that needs to be treated,” the doctor is showing the patient the radiograph with the cavity clearly circumscribed and labeled by the AI. The patients get a clearer understanding of what exactly is going on in their mouth, and that gives them greater confidence in the treatment recommendation. This is what dentists report to us, but I think it’s reasonable to extrapolate that the better patient communication that the AI enables is leading to greater patient trust — and hopefully improved patient retention.
Now that we’re in more practices, we’re developing research looking at real-world impact to verify anecdotal accounts of patient perspectives. We’re starting that research in Germany with academic support. There are many questions we’d like to answer over time. Does AI help speed up patient visits? Do patients trust doctors who use AI more than doctors who do not? Do they accept treatment from AI-equipped doctors at a higher rate? We should have answers to some of the questions pretty soon.
GN: When you think of dentistry in 10-15 years, how will technology have changed the profession and patient experience?
Ophir Tanz: I expect most dental offices in the world will be applying AI in some form — and often across much of the practice workflow, both clinically and operationally. Charting, scheduling, inventory management — these kinds of tasks will be accomplished with markedly more efficiency than they are today. The time gained should deliver some combination of the following benefits: lower costs of care, more patient volume and high-quality patient-doctor interaction. From a clinical perspective, we’ll have a higher standard of patient care across the board and better population-wide oral health.
At the farther end of that timeframe, I hope we’ll see AI facilitating more predictive and preventative dental care. It is not unreasonable to anticipate that we will be bringing a wide array of data points from outside of the patient’s mouth — medical records, family history, daily habits and lifestyle information — to bear both in developing individualized courses of treatment and in establishing the kinds of oral-systemic health links that have proved so hard to pin down to-date. As I noted previously, we see dentists more frequently than we do any other doctor — so it would be a wonderful thing if AI could give us insights that transform the mouth into a window to our heart, lungs or brain. That future may be more than 15 years out — but whenever we reach it, we’ll have AI to thank.