New modes of artificial intelligence help dentists

Search Dental Tribune

“AI’s primary function is to bolster human skills, not to overshadow them”

A variety of AI technologies can help dentists with everything from running their practice, planning a surgery, making a diagnosis to predicting disease. (Image: Antonio Marca/Shutterstock)

Artificial intelligence (AI) presents dental clinicians with a multitude of helpful aids, and researchers are continually expanding the capabilities that AI can offer a busy clinic. Dental Tribune International spoke with Dr Hanyao Huang, a researcher in the Department of Oral and Maxillofacial Surgery at West China Hospital of Stomatology at Sichuan University in Chengdu in China, regarding his research into the application of AI in dentistry.

Dr Hanyao Huang. (Hanyao Huang)

Dr Huang, could you kindly share with our readers your motivations for researching ChatGPT’s application in dentistry and explain why you feel they should be excited about using AI in their practice and in the field of dentistry?
The journey of AI in dentistry has been progressive. Traditional AI applications, like computer vision, have been instrumental in enhancing diagnostic precision. These systems primarily revolved around data extraction, and the decision-making was largely reliant on human expertise. However, the advent of large language models (LLMs) like ChatGPT has ushered in a paradigm shift. Instead of just processing and presenting data, LLMs facilitate dynamic conversations, tapping into vast reservoirs of human knowledge across multiple sources. This not only aids in information dissemination, but also fosters a more interactive and holistic approach to problem-solving, bridging the gap between raw data and human-centred decision-making in the dental domain. Now, many studies that were previously deemed infeasible can actually be carried out.

Your study elaborated on how AI and LLMs are becoming more valuable in medical and dental diagnostics. In the study, you also provided a really clear example of how to use ChatGPT in a manner that effectively creates a consultation on treatment options. What tips would you offer to a dentist who wants to try out natural language reasoning (NLR)?
Computer vision techniques in dentistry primarily focus on analysing images from CBCT and radiographs to detect dental issues and assist in surgery, offering quantitative insights from radiographs and real-time imaging. We utilise it in order to decide what to do next. In contrast, LLMs like ChatGPT, while not processing images directly, excel in understanding and generating text, which aids in patient interactions, research and education. Integrating both can provide a holistic approach, in which after image analysis, LLMs can generate detailed reports or explanations, combining the strengths of both visual data extraction and textual understanding. Inexperienced dentists may be unaware of certain information, and they can be assisted by LLMs and NLR. However, as this technique is still under development, dentists cannot rely on it fully at present. LLMs can gather abundant data, but the sources are not all reliable. Further studies should be undertaken that focus on sorting and evaluating the authority of the available data.

A visual question answering example framework with the assistance of an image encoder, which generates a natural language representation from a radiograph. (Image: © 2023 Huang et al., licensed under CC BY 4.0)

You provided an example of using NLR to help dentists lower the likelihood of potential adverse drug reactions in relation to dental procedures and other drugs. Could you describe what it might look like when a dentist chooses to use ChatGPT or other forms of AI to make their practice and treatments as effective as possible?
It would be very effective. Of course, as I mentioned in the last question, strict and proper data sorting is extremely important before this application. After that, for dentists, LLMs will help provide a more careful approach to treatment and help decrease the incidence of complications. For example, after the patient has described the chief complaint, medical history, history of allergies, medication history and diet history should be assessed, and this could be done with LLMs like ChatGPT. Then the programs can identify the more critical points or risk factors and better prepare the treatment plan. This would be especially helpful for young and inexperienced dentists.

“It is crucial to view AI as an enhancer rather than a substitute.”

One of the major roles of AI in a clinic is visual data generation and analysis, but the idea of a program being a better diagnostician than a human makes some clinicians wary that they could be out of a job. What reassurances would you offer someone who has heard rumours about AI in the medical setting?
The incorporation of AI in medical settings, particularly in clinics, marks a significant shift. However, it is crucial to view AI as an enhancer rather than a substitute. AI’s primary function is to bolster human skills, not to overshadow them. AI excels at swiftly processing vast amounts of data and pinpointing patterns that might elude human observation, yet clinicians should have the final say. Even though AI can sift through data and propose diagnoses, many medical choices demand a comprehensive grasp of the patient’s background and other relevant details. AI supplies data, but intricate decisions typically hinge on human discernment. The introduction of AI in clinics paves the way for ongoing education. Medical professionals can harness AI to keep abreast of current studies, unique cases and emerging methodologies. In the same way that technology has reshaped various professions, the role of medical practitioners may also adapt with AI’s inclusion. This evolution does not signify job elimination but rather a realignment of duties, fostering enhanced patient care. To sum up, AI holds immense promise in refining diagnostic precision and speed, but it does not supplant human medical professionals. When utilised aptly, it can amplify human potential and improve healthcare results.

Could you please tell our readers something more about your study or about other ongoing projects you are excited about?
We are embarking on an ambitious project to fine-tune an offline LLM with knowledge curated from guidebooks across all specialties. Our goal is to create a more accurate and context-aware chatbot tool that can assist in various domains. We encourage all researchers and clinicians to join this work.

Editorial note:

The study, titled “ChatGPT for shaping the future of dentistry: The potential of multi-modal large language model”, was published online on 28 July 2023 in the International Journal of Oral Science.

To post a reply please login or register