How Is AI Used in the Dental Sector?

Artificial Intelligence (AI) is developing at a fast pace nowadays and is expanding especially to the various health fields such as cardiology, oncology, Medical Imaging, dermatology, odontology, or even radiology. It worries as much as it inspires, and we can no longer pretend that it does not exist. Like all new technology, artificial intelligence holds enormous potential for improving the health of millions of people around the world, but like all technology it can also be misused and cause harm,” said Dr Tedros Adhanom Ghebreyesus, WHO Director-General when introducing the first global report on Artificial Intelligence (AI) in health Ethics and governance of artificial intelligence for health.

When it comes to dentistry, AI has a variety of applications, from a simple tool that makes a dentist's day easier to one that makes diagnoses. So should we dentists be afraid of AI?

Let us first look at what AI is primarily used for today: improving practice productivity by reducing time-consuming, repetitive tasks.

Many AI-based software products focus on very specific tasks. For example, AI is used in voice commands like DEXvoice, the "Alexa" for dentists from Simplifeye, and DEXIS. This software can extract x-rays, patient records, and charts with a simple voice command. This solution provides hands-free access and can ultimately speed up the dentist's work. In particular, Spotimplant offers an AI-based implant identification service.

Also to increase productivity, Montreal's CHUM is currently experimenting with emergency room triage software. Patients arrive at the ER, input their information into a computer, which then sorts them according to their level of urgency. The AI also determines whether the problem is respiratory, pulmonary, cardiac, or other. "We are currently comparing this triage performed by the machine to a human triage. The machine saves time, but we want to make sure that this triage is done wisely and is of good quality," says Dr. Fabrice Brunet, president, and CEO of CHUM, who acknowledges the importance of human oversight versus an AI, which seems to be beneficial.

Other AIs go further: they do not limit themselves to a simple gain in productivity, but go so far as to help the user make a decision. Herein lies the crux of the ethical problem of AI and the questions it raises.

The goal is to improve patient care and make a diagnosis when appropriate. To know if a mole is a melanoma, to interpret a brain scan or a lung x-ray: AI can offer the doctor a second opinion or even detect very subtle abnormalities to make a definite diagnosis or suspect pathologies. The most sophisticated technology is image recognition through machine learning. By feeding billions of data already diagnosed by experts into a learning program, the algorithm learns to recognize a pathological sign on a digital image. For example, Google AI has developed an AI that diagnoses lung cancer with a 94.4% success rate (1). This can also avoid invasive tests like biopsies. And of course, AI only provides a second opinion that the doctor may - or may not - consider to make their own and unique final diagnosis. So AI is just a tool for the practitioner.

But what about when the patient comes in for a routine visit? VisualDx, another program in the medical field and soon to be in the dental field, allows doctors to input patient images and symptoms and instantly generate a list of possible diagnoses. This gives the doctor an overview of all possible pathologies. The AI then only suggests scenarios and completes the dentist's view at best.

Instead of suggesting several probable diagnoses, Evidentiae's algorithm (2) focuses on making only one and as accurate as possible. It is designed to extract information from medical and dental histories and map test results to create a comprehensive overview of your patient's dental health. It develops a thorough diagnostic opinion for periodontal concerns, biomechanical parameters, functional decisions, and dentofacial changes.

AI also helps with prevention, for example, by identifying an allergy risk or drug interaction that a doctor might not have thought of.

AI also helps labs in their research activities to predict the future. It takes about ten years and millions of dollars to bring a drug to market. And in the case of epidemics like Covid-19, the need can be urgent. One way to shorten the time it takes to develop a vaccine is to optimize preclinical research. That's the goal of InVivo AI, a startup founded by three Quebec Ph.D. students. "Currently, the drug discovery process is quite intuitive," explains Terence Bois, co-founder of InVivo AI. "For a given therapeutic target, a researcher tests several molecules, often in a rather random way, and repeats the experiments until he finds one that is active for the target of interest, all in a very iterative way. InVivo AI's technologies analyze the data generated by these researchers and create models that allow them to computationally simulate these experiments and go through this process faster."

For example, an artificial intelligence (AI) network developed by Google's AI subsidiary DeepMind has taken a big step toward solving one of biology's biggest challenges: determining the 3D shape of a protein from its amino acid sequence. "In a sense, the problem is solved," says John Molt, a computational biologist at the University of Maryland College Park, the co-founder of CASP, a global, community-based experiment in protein structure prediction that has taken place every two years since 1994.

AI is also a highly effective communication tool. Patient understanding and education are key to adherence to the treatment plan. Did you know that 56% of your patients do not accept their treatment plan because they do not understand the importance of seeking treatment and that 67% of patients cannot locate their pathologies on their dental X-rays? Some AI-based software, such as Allisone, allows you to highlight features on patients' X-rays to explain dental X-rays to your patients in a visual, fun and educational way, making it easier for them to understand and more involved in their oral health.

So can we trust AI? What are the ethical implications of this?

Clearly, in all of the above cases, AI can only be a tool relied upon by the practitioner. Unless AI is used to present information about the patient, at best it will only suggest a possible diagnosis. AI never makes a decision without the practitioner confirming that opinion. We believe this is exactly how AI should be used: as a "bonus", a great opportunity to reduce errors and oversights, a suggestion for the doctor's opinion. Why pass up such a medical advance with proven positive results? "AI has great potential to strengthen the effectiveness and efficiency of our healthcare system. The first ethical risk would be to close ourselves off to innovation and not open ourselves up to this potential for new solutions." David Gruson , Board Member of Sciences Po Paris Health Chair, Founder of Ethics- IA in an interview with UFSBD, the French Union for Oral Health.

Especially since AI will be drastically controlled if it helps with diagnosis. In fact, the new EU Directive 2017/745 now requires AIs that are considered medical devices to have class II -A certification. Self-certification, which was recommended for Class I, is therefore now replaced by control by a third-party organization and the performance of clinical studies to validate the performance of the system. The aim is to improve the integrity of computerized medical devices, their operational safety and the functional quality that enables them to make a diagnosis.

The thorniest issue remains the dreaded image of robotic AI replacing dentists.

Indeed, the image of the robotic dentist Yomi has been going around the world, but deserves to be clarified. Yomi is actually a new robot capable of placing dental implants, according to a study published by the journal South China Morning Post. In China, two new 3D-printed teeth were successfully implanted in a woman's mouth with the help of Yomi. But first of all, this technology is nothing without humans, dentists programmed it, and only after the human thought process of planning before implantation can Yomi act, like a technological arm that was previously remotely controlled. Only when the bone has been studied in its height and thickness and the operation has been prepared can it act, elements that are far too difficult for robots to manage. It then responds to simple commands like a computer, nothing more. So it is not AI and is much closer to the simple tools that have long been used by practitioners. So robotic dentists are not for tomorrow.

Dr. Torsten Meyer-Elmenhorst, dentist at Ivoclar Vivadent AG , is not worried about these developments, "Robots are only capable of doing what humans teach them to do. They will perform highly precise steps based on specific data sets. Nevertheless, such systems will always play a supporting role. The knowledge of qualified dentists - and therefore human specialists - will remain indispensable for making indications and decisions in the future."

Marie Christine Jaulent, research director at Inserm Medical Informatics and Knowledge Engineering Laboratory for e-health, also stresses the lack of reasoning ability of robots, which therefore cannot replace doctors(3). Only their incredible memory can support them.

We are now more aware of how AI can benefit us, that it is already being used in the world of health and dentistry, and that it is just a tool for the dentist to make his job easier and enable him to make an even better diagnosis.

1. “Santé et intelligence artificielle : quelle révolution nous attend ?”, Céline Deluzarche, FuturaSanté, 17 Octobre 2020.
2. “Pourquoi l’intelligence artificielle est l’avenir de la dentisterie”, Dr. Marc Cooper, DentisFuturis, 26 Novembre 2017.
3. Vidéo Youtube “Les médecins de demain remplacés par des robots... Vraiment ?”, publiée par la chaîne Inserm, 25 janvier 2019.
4. “Impacts du nouveau règlement 2017/745 sur la gestion biomédicale des dispositifs médicaux”, V. Boissart, NCBI, 31 mars 2021.