Generative Pre-trained Transformer: Trends, Applications, Strengths and Challenges in Dentistry: A Systematic Review
Article information
Abstract
Objectives
The integration of large language models (LLMs), particularly those based on the generative pre-trained transformer (GPT) architecture, has begun to revolutionize various fields, including dentistry. Despite these promising applications, the use of GPT in dentistry presents several challenges. Ongoing research and the development of robust ethical frameworks are essential to mitigate these issues and enhance the responsible deployment of GPT technologies in clinical settings. Hence, this systematic review aims to explore the trends, applications, strengths, and challenges associated with the use of GPT in dentistry.
Methods
Articles were selected if they contained detailed information on the application of GPT in dentistry. The search strategy used in systematic reviews follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Our search of databases and other sources yielded a total of 704 studies. After removing duplicates and conducting a full-text screening, 16 articles were included in the review. The methodological quality of the research was evaluated using the Critical Appraisal Skills Programme (CASP) checklist.
Results
Out of a total of 91 articles published on GPT in dentistry, 20 were editorials and 11 were narrative reviews; these were excluded, leaving 60 original research articles for further analysis. The articles were assessed based on the type of results they provided. Ultimately, 16 articles that reported positive findings with robust methodology were included in this review
Conclusions
The results highlight mixed responses; therefore, further research on integration into clinical workflows must be conducted with extensive methodological rigor.
I. Introduction
“Artificial intelligence (AI) encompasses a wide-ranging domain within computer science that focuses on creating intelligent machines capable of executing tasks usually requiring human cognitive abilities” [1]. The field of AI encompasses a wide range of disciplines dedicated to developing intelligent machines capable of performing tasks that typically require human cognitive abilities [1]. The adoption of AI and related technologies is increasingly common in both business and society, and it is beginning to transform the healthcare sector [2]. AI frameworks and platforms simplify access to this technology by standardizing the algorithm development process, thereby enabling researchers across disciplines to concentrate on designing neural network structures and ultimately delivering improved solutions to their respective challenges [3]. Natural language processing (NLP) is a discipline that combines computer science, AI, and computational linguistics to enable computers to process and understand human language texts. NLP empowers programs to analyze unstructured text data, which is particularly useful in the healthcare sector for managing information overload—for example, aggregating and summarizing patient notes, analyzing treatments, extracting data from extensive discharge summaries, and interpreting patient inquiries [4]. The applications of NLP in medicine are diverse.
The integration of large language models (LLMs), particularly those based on the generative pre-trained transformer (GPT) framework, is beginning to transform multiple domains, dentistry included. Among the various LLM technologies, GPT-based tools such as OpenAI’s ChatGPT have emerged as effective resources capable of generating humanlike text and facilitating complex interactions. AI systems, such as GPT-4, have demonstrated remarkable potential in natural language processing [5].
The integration of GPT in dentistry can streamline communication, improve patient education, and assist in clinical decision-making, thereby transforming the landscape of dental care. One of the primary applications of GPT in dentistry is in patient communication. By generating personalized educational materials, GPT helps dental professionals convey complex information in a more accessible manner. This is particularly important in pediatric dentistry, where effective communication is crucial to ensure that both children and their guardians understand treatment options and oral health practices [6]. Furthermore, GPT can assist in automating routine inquiries, allowing dental staff to focus on more complex patient needs.
In addition to enhancing communication, GPT has the potential to support clinical decision-making. By analyzing patient data and generating evidence-based recommendations, these models can aid dentists in diagnosing conditions and formulating treatment plans [7]. This capability is especially valuable in a field where timely and accurate information is essential for effective patient care. Research indicates that LLMs like GPT can significantly improve diagnostic accuracy and treatment outcomes, making them a promising tool for clinical applications in dentistry [8]. The potential for GPT to reduce the workload of dental practitioners while enhancing patient engagement and satisfaction is significant [9].
Despite these promising applications, the use of GPT in dentistry presents several challenges. Concerns regarding the accuracy of AI-generated information, ethical implications, and the risk of misinformation remain prevalent [9].
Studies have shown that although large language models excel in knowledge-based tests, their accuracy in medical and dental contexts may be inadequate [10]. Hence, this systematic review aimed to explore the trends, applications, strengths, and challenges associated with the use of GPT Transformers in dentistry. By synthesizing existing literature, this review provides a comprehensive overview of how GPT technology can be effectively integrated into dental practice while addressing its limitations and ethical considerations. The findings will serve as a foundation for future research and will guide dental professionals in leveraging AI tools to enhance patient outcomes and overall care quality.
II. Methods
In recent years, GPT has been introduced into dental practice as a potential solution to various challenges. This is a relatively new field of research in dentistry and has not received as much attention as other areas of AI. This review followed the guidelines provided by the Indian Council of Medical Research (ICMR) in the “Beginner’s Guide for Systematic Reviews: A Step-by-Step Guide to Conducting Systematic Reviews and Meta-analyses.” This systematic review aimed to address the following research questions:
What types of GPT are being utilized in various dental practices?
Which article types and specific areas of dentistry benefit most from GPT integration?
What are the key advantages of using chatbots? and
What limitations exist in adopting this technology in dentistry?
Using the SPIDER tool, the research question framework was defined as follows:
- Sample: All original studies reporting on GPT applications in dentistry, published without date restrictions;
- Phenomenon of interest: Dentistry;
- Design: Observational studies;
- Evaluation: Trends, applications, merits, challenges, and potential future research areas; and
-Research type: Qualitative.
The systematic review was registered (ID: CRD42024599702) on PROSPERO (https://www.crd.york.ac.uk) and was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Since this study did not involve human subjects, no ethical clearance was required.
A comprehensive search was performed across several bibliographic databases including ScienceDirect, PubMed, Scopus, EBSCO, IEEE, as well as grey literature sources (Open-Grey). Since the literature contains varied terminologies for GPT, the search strategy utilized the following keywords and Boolean operators: (“Generative Pre-Trained Transformer” OR “GPT” OR “Large Language Models” OR “LLMs” OR “Natural Language Processing” OR “Chatbot” OR “conversational agent” OR “Virtual assistants”) AND (“dentistry” OR “oral health” OR “dental practice” OR “dental education” OR “dental diagnostics”). The search period spanned from 2018 (the inception of GPT) to 2024.
Inclusion criteria are defined as the articles that contained detailed information on the application of GPT in dentistry; and only original research articles were considered. Exclusion criteria are defined as the articles not written in English, lacking full-text access, or whose content was not specifically focused on GPT or dentistry. The review excluded research articles that used ChatGPT without evaluating its advantages and disadvantages; systematic reviews, narrative reviews, and editorials (including letters to the editor, guest editorials, opinion pieces, insights, short communications, commentaries, and perspectives).
Given the significance of chatbot interaction and development to our study, the aforementioned databases were systematically searched for relevant publications. The search yielded a total of 704 articles across various sources: ScienceDirect (210), PubMed (61), Scopus (288), EBSCO (138), IEEE Xplore (2), and other records (5). The search strategy employed in this systematic review is illustrated in Figure 1 according to the PRISMA guidelines. Initially, using Rayyan software (https://www.rayyan.ai/), only publications that met the inclusion and exclusion criteria were selected. The selected articles were imported from the databases, and two independent, blinded researchers conducted the data extraction process to minimize bias and improve reliability. Any discrepancies were resolved through discussions and consensus meetings, with a third researcher consulted when necessary. The articles included were then exported for further analysis or reporting. After eliminating duplicates, the initial screening yielded 91 articles on the application of various GPT models in dentistry. Figure 2 illustrates the publication trends among the 91 articles published between 2018 and the present. Of these 91 articles, 20 were editorials and 11 were narrative reviews, which were subsequently excluded. The remaining 60 original research articles underwent an additional screening process. The articles were assessed based on the type of results they provided. Studies that presented both positive and negative findings were classified as neutral. The studies were categorized as having a positive, neutral, or negative stance toward the use of GPT in dentistry, as depicted in Figure 3. Out of the 29 articles that reported positive results, 16 studies with robust methodology were ultimately included in this review.
Search strategy used in this systematic review.
Publication trends.
Type of results obtained in original research.
Each study was thoroughly reviewed, with each focusing on a single category. Studies investigating ChatGPT’s role in identifying oral health conditions—by analyzing patient symptoms, clinical records, or imaging data—were classified under clinical diagnosis. These studies assessed ChatGPT’s ability to serve as a diagnostic aid by enhancing the detection of conditions such as dental caries, periodontal disease, or oral cancer. Research examining ChatGPT’s role in supporting dentists with treatment planning and management was categorized as clinical decision-making. These studies evaluated its potential to suggest evidence-based interventions, prioritize treatment options, or simulate patient scenarios to inform decisions. Other studies investigated the integration of ChatGPT in oral health research methodologies, such as generating research ideas, drafting protocols, conducting systematic reviews, or analyzing trends in patient data. The category of dental education explored ChatGPT’s utility as a teaching aid for dental students and professionals. These studies involved its use for interactive learning, case-based simulations, or as a supplemental resource for understanding complex concepts. The methodological quality of the research was evaluated using the Critical Appraisal Skills Programme (CASP) checklist, which gathered essential information from each study.
III. Results
The results address the research questions posed in the review. All the studies had distinct objectives, employed suitable qualitative techniques, utilized appropriate study designs, addressed ethical considerations, and provided clear statements regarding their findings. Study characteristics, such as authors, publication year, country of origin, research design, and study setting, were documented. Key findings include major themes, types of GPT, and insights related to GPT usage, impacts, recommendations for future use, and limitations in dentistry. A total of 16 original research articles were analyzed across various fields of GPT application (clinical diagnosis, clinical decision-making, oral health research, and dental education). No randomized controlled trials have been conducted in this field. Clinical decision-making studies encompassed areas such as oral radiology, endodontics, oral surgery, orthodontics, and periodontics. Table 1 summarizes the observations from the included articles, while Table 2 presents the quality assessment based on the CASP Checklist.
Observations of the original research included in the review
Critical Appraisal Skills Programme Checklist for the included study
IV. Discussion
The integration of AI in healthcare has revolutionized numerous industries, with dentistry emerging as a notable beneficiary. Among the latest developments in AI, GPT stands out as a revolutionary technology. This technology enables dentists to achieve diagnostic accuracy, tailor treatments, and enhance educational experiences. This systematic review examines the trends, applications, developments, and challenges of GPT—particularly models such as ChatGPT—in dentistry.
1. Clinical Diagnosis
Recent studies have underscored the variable accuracy and precision of AI-powered chatbots relative to human performance. GPT can significantly improve diagnostic accuracy [8]. A study evaluating ChatGPT’s diagnostic accuracy in periodontitis [11] found that the model achieved 40% and 62% accuracy in complex dental examination challenges when generating differential diagnoses. However, when the differential diagnoses were benchmarked against the literature, accuracy improved markedly to 70% for ChatGPT-3.5 and 80% for ChatGPT-4. Although ChatGPT can offer valuable insights, its performance is heavily dependent on the quality of input data.
In a pilot study focusing on the diagnosis of oral lesions [12,13], ChatGPT demonstrated accuracy in answering questions using its reasoning capabilities. Its ability to analyze text inputs enables it to combine clinical observations with patient history, which is essential for accurate diagnosis.
Conversely, another study revealed that ChatGPT’s performance in endodontics was suboptimal, with only 30% of its responses deemed accurate when presented with clinical scenarios [14]. This suggests that while GPT models can provide useful insights, their accuracy remains limited in more complex clinical environments.
2. Clinical Decision-Making
Content-aware chatbots can support clinicians by acting as decision-making aids, particularly in implementing cone beam computed tomography (CBCT) guidelines in clinical practice [15]. Similar positive results were discussed in other studies [16]. However, other studies observed contradictory findings, reporting that ChatGPT failed to achieve consistent performance [17]. Although AI applications show promise, they have not yet reached the desired levels of accuracy and consistency. These studies emphasize that ChatGPT in its current form should not be used without caution [18]. This implies that careful attention and continuous assessments are necessary to ensure precision in clinical decision-making.
3. Oral Health Research
GPT-4 offers numerous suggestions for researchers encountering challenges in identifying new topics; however, it is important to note that these suggestions are not original and require further exploration [19]. Nonetheless, its efficacy is limited by its inability to critically evaluate literature, distinguish between predatory and reputable journals, and maintain scientific accuracy and reliability [20]. Future updates for ChatGPT ought to prioritize minimizing false positives and false negatives to improve the overall effectiveness of the system [21]. Moreover, while AI is at the forefront of many scientific domains, using it without proper verification can lead to inaccurate and unethical outcomes, as AI-generated content does not analyze information in the same manner as a human researcher or writer [22].
4. Dental Education
The ChatGPT Transformer offers an intelligent learning platform that supports student education by customizing learning materials to individual needs [23]. It provides personalized learning experiences tailored to each student’s unique requirements. GPT can help students understand complex dental concepts through interactive materials and immediate feedback. For example, students can interact with GPT to clarify doubts, participate in interactive lectures, and enhance their problem-solving skills in a supportive environment. Furthermore, GPT assists in scientific writing by helping learners organize their essays and suggesting related materials [24–27]. Access to online educational materials has enabled thousands of students to benefit from effective and timely resource utilization [28]. However, further studies are required to assess its overall effectiveness [29].
Despite its benefits, GPT in dentistry has notable drawbacks. The most concerning limitation is the accuracy of the information it produces. Additionally, the ease with which GPT structures content raises concerns about academic dishonesty, as students may misuse it for plagiarism; comprehensive academic policies are necessary to mitigate these risks. Furthermore, overreliance on GPT may hamper the development of critical thinking and problem-solving skills, as students might favor AI-generated solutions over thorough engagement with the material [30].
Considering the observations from the studies included in this review, the findings reveal mixed responses. ChatGPT has significant potential to revolutionize dentistry by offering various applications and benefits. Nonetheless, researchers and academics must exercise caution when using ChatGPT, as AI-generated content cannot replicate the analytical depth of a human author or researcher. Thus, it is concluded that the application of GPT in various areas of dentistry remains a subject of debate. The tables in this manuscript outline various recommendations and future research avenues for further exploration of this technology in dentistry. Therefore, future research into integrating GPT into clinical workflows should be conducted with stringent methodological rigor.
Notes
Conflict of Interest
No potential conflict of interest relevant to this article was reported.