
17 minute read
Tech Tool: Artificial Intelligence Chatbots Play Increasing Role in Dentistry
By William S. Bike
In general, the concept of artificial intelligence (AI) is using software to execute tasks formerly performed exclusively by humans.(1) AI chatbots, the most prominent of which at this time is ChatGPT (which stands for “Chat Generative Pre-Trained Transformer”), use AI software to train on large language models — vast amounts of textual data from various diverse sources — in an attempt to both understand conversational questions and generate humanlike responses.(2) AI chatbots have begun to play a huge role in numerous sectors of society, including healthcare and dentistry. In dentistry, AI and ChatGPT are becoming integrated in three key areas: education, practice management and diagnosis. While the technology’s advantages are promising, dentists and legislators alike are cautious that standards need to be put in place before even more widespread adoption.
Education
“Educators are incorporating generative AI technology in dental schools,” explained Kyle Stanley, DDS, co-founder of Pearl, a company that uses AI to enhance patient care in dentistry. “AI can simulate complex dental procedures, provide virtual patient interactions, and offer personalized learning modules that adapt to each student’s pace and understanding.”
Cortino Sukotjo, DDS, PhD, MMSc, MHPE, chair of the department of prosthodontics at the University of Pittsburgh School of Dental Medicine, said, “Generative AI technology, like ChatGPT, is being utilized in dental education to create more interactive and personalized learning experiences.”
ChatGPT is an advanced conversational AI model developed by OpenAI.(2) It is available through a web interface available to both the public and professionals, and it’s described as easy to use.(3)
Sukotjo was one of the authors of a study published recently in The Journal of Prosthetic Dentistry in which dental school faculty deployed an experimental chatbot.(4) “It was designed to respond to clinical procedure questions frequently asked by students,” he said. “The chatbot reduces the need for faculty intervention, thereby decreasing delays and bottlenecks in the clinic. Students found the chatbot to improve their clinical experience while reducing anxiety.”
A chatbot can answer student questions such as what instruments they will need for a procedure or which code is appropriate for logging information in the electronic patient record.(5) Unlike faculty, the chatbot is available 24 hours per day, seven days per week.
“Future usage will include computer vision helping students identify carious lesions and periodontal disease utilizing 2D radiographs,” said Callan D. White, DDS, FAGD, of Asheville, North Carolina, who has researched dental AI for AGD’s Dental Practice Council as chair of its Subcommittee on AI and Augmented Intelligence.
AI chatbots are being used not only for student education, but also for patient education. “There are really two scenarios,” explained Kaveh Nedamat, DDS, MBA, founder and CEO of Toothly AI in Toronto, Ontario. Toothly is a platform that supports dentists in providing faster and more accurate diagnoses.
“The first scenario is where a dentist uses ChatGPT within the appointment to provide support in their communication with patients,” Nedamat said. The second is in patients’ search for information via search engines outside of the appointment. “There is often misinformation with this type of search,” he added, noting that patients “lack context and often process information in a way that is not helpful.”
Often patients don’t have the time, ability, knowledge, etc., to understand what is and isn’t rooted in science,” he added. “I would say there is also a predatory element out there trying to spread misinformation in order to sell prospective patients products that make great claims. Examples would be misinformation surrounding fluoride, amalgams, root canals, etc.”
Toothly, Nedamat noted, is working to develop a solution that bridges this knowledge gap.
“ChatGPT is being used to provide patients with easily understandable information about dental procedures, oral hygiene and preventive care,” Stanley said. “Informed patients are more likely to accept recommended treatment plans.”
“ChatGPT and similar AI tools are revolutionizing patient education by providing easily accessible, understandable information about dental health,” said Qiao Fang, DDS, MSD, visiting clinical assistant professor in the department of restorative dentistry at the University of Illinois Chicago College of Dentistry, and another author of The Journal of Prosthetic Dentistry study.
White agreed. “Patients are educating themselves with the use of large language models (LLM) like ChatGPT,” he said. LLMs understand, process and generate human language and information based on large amounts of data.
Dental researchers are harnessing ChatGPT to analyze data and dental literature, identify patterns and generate insights.(2) “Dental researchers today have an amazing opportunity to utilize LLMs like ChatGPT to scan thousands of studies and papers,” White said.
“ChatGPT can quickly sift through electronic health records, research papers and clinical trial data to identify patterns and correlations that might be missed by human researchers,” Fang said.
White explained that ChatGPT’s “prompts are crucial in selecting the correct information needed and can save researchers a tremendous amount of time.”
“Dental researchers are leveraging ChatGPT to sift through large datasets and extract valuable insights and patterns,” Stanley said. “This helps in identifying trends, understanding patient outcomes, and conducting epidemiological studies with greater efficiency and accuracy.”
“ChatGPT has demonstrated its effectiveness in helping scholars with the authoring of scientific research and dental studies,” Anushree Tiwari et al. stated in Cureus, because researchers “have been able to summarise, interpret and rephrase scientific data by using ChatGPT.”(6)
Practice Management
Concerning practice management, “with AI providing the capacity for supreme efficiency in dentistry,” Nedamat said, “it can be applied to dental practices, especially in settings where resources are limited and need to be optimized. I view computer vision-assisted diagnostic support as part of the solution to health deserts.”
“AI chatbots are being integrated into office management software to handle appointment scheduling, patient inquiries and reminders,” Stanley said, “which reduces the administrative burden on dental staff, improves patient communication and enhances overall office efficiency.”
Edward J. Zuckerberg, DDS, FAGD, works with several companies involved in AI and explained that AI can be used “to automate the creation of SOAP [subjective, objective, assessment and plan] notes from verbal conversations between dentists and patients and to incorporate these into the practice management software.”
“AI chatbots help reduce the administrative burden on staff, allowing them to focus more on patient care,” Sukotjo said.
Fang believes AI “has significantly boosted the popularity of teledentistry,” he said. “AI-powered tools enable more effective remote consultations by assisting dentists in diagnosing conditions based on patient-provided data and images. These tools can also triage cases to determine which ones require in-person visits, thus optimizing the use of clinical resources.”
Stanley noted that “AI-driven tools can assess patient photos and radiographs submitted online, allowing dentists to provide preliminary diagnoses and treatment recommendations without an in-person visit.”
“[AI] can mean a dentist in a central location can manage multiple hygienist-driven practices at once,” Nedamat explained. “This is done via video chat and can be made more efficient with the assistance of AI-driven diagnostic support. The dentist and hygienist can effectively perform a clinical exam, communicate the results in a visually appealing way to the patient and triage the patient to the appropriate dental specialist. Those patients residing in oral health deserts can still receive excellent care and only travel when necessary.”
Scott Froum et al., in Dental Economics, noted that ChatGPT can be used by the practice for website and search engine optimization (SEO) and development.(3) ChatGPT can also be used to reply to patient emails; create and send surveys; and write blogs, social media posts and newsletters.(7)
Revenue cycle management may also be improved, Stanley said. “AI helps streamline revenue cycle management by automating billing processes, predicting payment delays and identifying errors in claims submissions,” he said. “This ensures faster and more accurate reimbursements, reducing the time and effort required for manual processing.”
“This is an area of AI development that will be extremely beneficial for everyone involved in the dental care setting and is developing very quickly,” White added. “Not only will it help patients understand and better utilize their insurance benefits, but it will also help dental teams be more efficient in insurance administration. The goal is seamless communication between the patient, provider and insurance company where verification, adjudication and payment can happen in real time.”
“AI systems can quickly identify discrepancies or errors in billing and claims, reducing the time and effort required to resolve these issues,” Fang said.
Diagnosis
Concerning diagnosis, “AI deep learning algorithms can detect anomalies like caries, periodontal disease and other oral pathologies with high accuracy,” Fang said. “These tools provide a differential diagnosis, thereby enhancing the dentist’s diagnostic capabilities, ensuring nothing is overlooked and streamlining the screening process.”
Zuckerberg is chief dental officer for Viome, a Seattle-area company that has “the first noninvasive oral and throat cancer screening from saliva samples, which by itself is pretty amazing. But they also deliver six detailed oral health scores: gingival health, caries susceptibility, breath odor, pathobiont activity, fungal activity and genotoxic activity,” he explained. “They literally have to sift through trillions of nucleotide data points and weed out unwanted data to come up with the relevant scores. This task would be impossible without AI.”
In the office, Zuckerberg said, “For speed of analysis and thoroughness as it relates to some obscure information on the image that dentists might overlook — be it a small radiolucency or calcification that might be dismissed as insignificant — the AI will document these every time and measure changes over time in a manner that human eyes cannot.”
Zuckerberg is passionate about oral cancer detection/prevention and offered an example where AI can be helpful: “A lesion is nonmalignant with some dysplastic changes or conditions that fall under oral pre-malignant disease, like leukoplakia and others. The surgeon is put in a difficult position of either telling the patient they don’t have cancer but urging the importance of future monitoring with re-biopsy periodically, or the aggressive approach of total excision.”
Zuckerberg noted that Proteocyte AI in Toronto, for which he is a senior adviser, now has an AI tool that can calculate “a score, expressed as a percentage likelihood that a particular lesion will undergo a malignant transformation over the next five years,” giving “both the surgeon and the patient a realistic basis for a treatment choice.”
“Computer vision is a facet of AI that can aid dentists in the examination of radiographs,” White said. “These AI algorithms have learned to detect variations of contrast within radiographs to help dentists detect carious lesions. The AI can measure the size and depths of the lesions. There are also AI algorithms that can detect bone loss around teeth and subgingival calculus, helping providers diagnose periodontal disease.”
AI analysis of radiographs can identify caries and other abnormalities “with high accuracy,” Stanley noted. “Beyond radiographs, AI tools are also being used to analyze intraoral scans and patient records to assist in diagnosing conditions and planning treatments.”
Among diagnostic uses, Zuckerberg explained, are “AI utilization merging diagnostic information from cone beam computed tomography images to better enable endodontists to locate second mesiobuccal and accessory canal locations as well as blockages and extreme curvatures that could not be seen easily.” AI can also be used to “design prostheses, determine ideal types — removable versus fixed — and select components and materials.”
“AI is also used in orthodontics software to help visualize malocclusions and predict optimal treatment results for patients,” White added.
But AI isn’t just helping dentists streamline their processes. Dental insurers are also engaging the technology.
“Insurance companies are using AI to review and verify dentists’ diagnoses and treatment plans to ensure they align with coverage policies,” Stanley explained.
This has some dentists concerned.
Stanley noted insurance companies’ use of AI may “lead to disputes if AI assessments differ from those of the treating dentist, potentially affecting reimbursements.”
“The same algorithms that are used by dentists to improve their diagnostic ability and patient communication are also being used by insurance companies to deny claims,” Nedamat said.
Therefore, Fang said, “it is crucial for dental professionals to maintain detailed and accurate records to support their diagnoses and treatment plans.”
“The good news is that, in most states, you cannot deny a claim without a human, so insurance companies are using this technology to approve your claims so they don’t have to go to the expensive human consultants,” Stanley said, noting that if AI designates a procedure as nonapproved, “it gets flagged and routed to a human.” White agreed. “As far as I know, and what we are allowed to know, a human still has to review a nonapproved claim prior to the denial,” he said.
White also offered a positive, saying, “Insurance companies are using AI abilities to help eliminate fraud, waste and abuse by auditing the data collected from submitted insurance claims.” However, he noted that “it should be of the highest importance for organized dentistry, like AGD, to ensure that any claim denials are reviewed by a dental consultant. AI should only be used as an audit for the insurance company.”
There is also some concern among dentists that AI is aiding insurance companies in “takeback” claims, where insurance companies request refunds for previously paid claims.
Stanley offers a positive, saying, “AI can audit past claims more efficiently, potentially identifying errors or discrepancies that lead to such refund requests.”
“The clinical exam is imperative, and AI can be used as an adjunct to assist in diagnosis and not the other way around,” Nedamat said. “This almost feels like it should be mutually exclusive, meaning companies providing support to dentists to increase case acceptance of the right treatment shouldn’t be then providing the same algorithms to insurance companies to deny claims or even takeback payouts, namely because there is a lack of context.”
Developing Standards
Organized dentistry — along with organized medicine and federal and state governments — are working to implement standards and oversight on AI. An American Dental Association (ADA) standards workgroup created a white paper that provides an overview of how AI is currently being used in dentistry and how it might be used in the future. The ADA is developing standards to guide dentistry’s adoption of AI. White has been involved in those efforts as a representative of AGD.
AGD also created an AI policy. In AGD’s policy statement, White states: “The most crucial element of AGD’s policy is the position that AI should never supersede or replace the dental practitioner in clinical decisions or in any way erode the patient/practitioner relationship.”(8)
AGD’s policy promotes transparency in the development and use of AI systems and states that, when third-party payers utilize AI, the technology “should not unduly track and/or attempt to influence recommended treatment and [should] ensure equal reimbursement for providers without regard to the use of AI within the practice.”(8)
Federal lawmakers are also leery of healthcare decisions made solely by AI. Earlier this year, the Centers for Medicare and Medicaid Services (CMS) issued a final rule that Medicare Advantage plans cannot make medical necessity determinations using an algorithm or software that does not take into account individual circumstances. Reps. Jerrold Nadler (N.Y.) and Judy Chu (Calif.) and Sen. Elizabeth Warren (Mass.) went a step further and, in June, sent a letter to CMS urging it to increase oversight of AI and develop guardrails on the use of AI and algorithms to determine Medicare Advantage coverage for seniors. The letter points out that “several [Medicare Advantage] insurers are using unregulated algorithms and AI tools to determine when to cut off payment for patient treatments.”(9)
State lawmakers are also turning their attention to AI in healthcare, attempting to create guardrails that promote privacy, data integrity and transparency. In 2023, 11 states introduced legislation related to AI use in healthcare, up from three in 2022.(10) These laws echo the themes of the majority of the nascent standards — that AI cannot replace humans when it comes to decision-making and that patients must be informed when it is being used.
Some of these laws are already being tested. In July 2023, a group of patients sued Cigna Healthcare, alleging that the company’s doctors circumvented the legally required physician review process by using an AI algorithm to automatically deny thousands of claims.(11) The case is still early, but it demonstrates that AI’s role in healthcare is far from agreed upon.
AI’s Future in Dentistry
“AI is here to stay and will be extremely useful in many facets of dentistry,” White said.
Fang offered a caveat, however: “We need to be aware that AI tools do not replace dentists’ expertise. AI diagnostic results still require supervision by human professionals.”
White offered his own caveat: “My concern is how this information will be stored, used and protected. There should be more transparency in the usage of patient information by AI companies and insurance providers.” He also called for “AI standardization to prevent biases,” and further stated that “AI should only be used as a tool and should never come between the patient and provider. AI software should ensure that patient information is stored correctly, used only for the purpose intended, and protected from breaches and nefarious purposes.”
Nonetheless, “AI usage in dentistry is growing quickly,” White said. “It will be extremely important to have testing and validation standards to prevent unwanted biases and optimize specificity and sensitivity outcomes.”
Both Fang and Sukotjo see AI involvement in predictive analytics for preventive care as well as the integration of AI with other emerging technologies such as augmented and virtual reality and robotics.
Froum et al. noted in Dental Economics that by utilizing the AI powering ChatGPT, researchers can create protein-language models to study biological data and optimize established molecules to improve the efficacy of and to repurpose drugs.(3)
Fang and Sukotjo also see an increasing need for attention to ethical considerations to ensure that AI technologies are used responsibly and effectively.
De Souza et al. in General Dentistry also cited ethical concerns, noting that “integrating ChatGPT in dentistry can be highly beneficial, but it is crucial to address ethical considerations, accuracy and privacy concerns.”(2)
Nedamat offered a final warning: “Democratization of health information is inevitable, and we as a profession will either adapt or die,” he concluded. “AI will not replace dentists, but a dentist with AI will replace a dentist without. Don’t be replaced.”
William S. Bike is a freelance writer and editor based in Chicago. He is a former director of advancement communications for the University of Illinois Chicago College of Dentistry. To comment on this article, email impact@agd.org.
References
1. Dysart, Joe. “Artificial Intelligence and Dentistry: Mostly Blue Skies Ahead.” AGD Impact, vol. 50, no. 4, April 2022, pp. 12-17.
2. De Souza, Lucas, et al. “ChatGPT and Dentistry: A Step Toward the Future.” General Dentistry, vol. 72, no. 4, July/August 2024, pp. 72-77.
3. Froum, Scott, et al. “ChatGPT: Will It Revolutionize the Dental Industry?” Dental Economics, 13 April 2023, dentaleconomics.com/science-tech/article/14290255/chatgpt-will-it-revolutionize-the-dental-industry.
4. Fang, Qiao, et al. “Artificial Intelligence (AI)-Driven Dental Education: Exploring the Role of Chatbots in a Clinical Learning Environment.” The Journal of Prosthetic Dentistry, 21 April 2024, doi: 10.1016/ j.prosdent.2024.03.038.
5. Mitchum, Rob. “Can a Chatbot Help Educate Dentistry Students?” UIC Today, 15 May 2024, today.uic.edu/ can-a-chatbot-help-educate-dentistry-students.
6. Tiwari, Anushree, et al. “Implications of ChatGPT in Public Health Dentistry: A Systematic Review.” Cureus, 2023 Jun; 15(6) e40367.
7. Wilson, Matt. “Meet Your New Dental Communications Assistant: ChatGPT.” Dentistry IQ, 13 July 2023, dentistryiq.com/front-office/article/14296324/chatgpt-a-game-changer-in-dental-patient-communicationand-content-creation.
8. Kluck-Nygren, Cindy. “Dental Practice Advocacy Survey Results: AGD Members’ Use of Artificial Intelligence (AI)/Augmented Intelligence (AuI) in the Dental Practice.” Academy of General Dentistry, agd.org/docs/ default-source/default-document-library/agd-member-survey-re-ai-aui_2024-june30_ cd.pdf?sfvrsn=b519d4ed_2. Accessed 10 Oct. 2024.
9. “Reps. Nadler, Chu & Sen. Warren Lead Bicameral Letter to CMS Urging Oversight of Artificial Intelligence and Algorithms Used in Medicare Advantage Coverage Decisions.” Congressman Jerry Nadler, 25 June 2024, nadler.house.gov/news/documentsingle.aspx?DocumentID=396200
10. “Artificial Intelligence and Health Care: A Primer.” National Conference of State Legislatures, 20 Aug. 2024, ncsl.org/health/artificial-intelligence-health-care-a-primer?utm_source=national+conference+of+state +legislatures&utm_term=0_-2a71222d97-%5blist_email_id%5d&utm_campaign=2a71222d97-thisweek-aug-4&utm_medium=email.
11. Greenberg, David S. “Health Insurers Sued Over Use of Artificial Intelligence to Deny Medical Claims.” ArentFox Schiff LLP, 22 Dec. 2023, afslaw.com/perspectives/health-care-counsel-blog/health-insurers-suedover-use-artificial-intelligence-deny.