Abstract
Background
Artificial intelligence chatbots have the potential to inform and guide patients by providing human-like responses to questions about dental and maxillofacial prostheses. Information regarding the accuracy and qualifications of these responses is limited. This in-silico study aimed to evaluate the accuracy, quality, readability, understandability, and actionability of the responses from DeepSeek-R1, ChatGPT-o1, ChatGPT-4, and Dental GPT chatbots.
Methods
Four chatbots were queried about 35 of the most frequently asked patient questions about their prostheses. The accuracy, quality, understandability, and actionability of the responses were assessed by two prosthodontists using five-point Likert scale, Global Quality Score, and Patient Education Materials Assessment Tool for Printed Materials scales, respectively. Readability was scored using the Flesch-Kincaid Grade Level and Flesch Reading Ease. The agreement was assessed using the Cohen Kappa test. Differences between chatbots were analyzed using the Kruskal-Wallis test, one-way ANOVA, and post-hoc tests.
Results
Chatbots showed a significant difference in accuracy and readability (p <.05). Dental GPT recorded the highest accuracy score, whereas ChatGPT-4 had the lowest. DeepSeek-R1 performed best, while Dental GPT had the lowest performance in readability. Quality, understandability, actionability, and reader education scores did not show significant differences.
Conclusions
While accuracy may vary among chatbots, the domain-specific trained AI tool and ChatGPT-o1 demonstrated superior accuracy. Even if accuracy is high, misinformation in health care can have significant consequences. Enhancing the readability of the responses is essential, and chatbots should be chosen accordingly. The accuracy and readability of information from chatbots should be monitored for public health.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12903-025-06267-w.
Keywords: Generative artificial intelligence, Public health, Digital health, Patient education, Dental prosthesis, Maxillofacial prosthesis
Introduction
The proportion of individuals aged 60 and older is increasing [1], and they often experience missing teeth and surrounding tissue due to aging and inadequate oral care. The global edentulism rate in individuals aged 60 and over has been calculated at 22.7% [2]. Fixed or removable prostheses are utilized to replace missing teeth. Patients with oral and dental health issues may suffer from pain and aspiration pneumonia [3]. Furthermore, if missing teeth are not replaced with prostheses, nutrition may be impacted, leading to concerns about the overall health and quality of life of elderly patients [4, 5]. While diseases, environmental factors, and early smoking contribute to the risk of cancer, age remains the most significant risk factor, with seven out of eight cases of oral cancer occurring in individuals over fifty years old [6]. Intraoral or extraoral maxillofacial prostheses can restore defects in the maxillofacial region resulting from cancer surgery or trauma. These prostheses are customized and made from various materials, depending on the patient’s needs and the location of the defect. Proper care and maintenance of the prostheses and awareness of when to consult a doctor are crucial for meeting functional and aesthetic requirements and ensuring longevity. While dentists can provide basic information, geriatric patients may struggle to reach healthcare professionals promptly due to limited physical mobility or reliance on caregivers. A British national survey indicates that individuals over 65 have the highest non-attendance rate at annual dental visits [7]. Accurate preventive oral health advice must be readily accessible to older adults, their families, and caregivers to help prevent the development or progression of oral health issues. However, older individuals who do not receive social care or visit a dentist regularly face significant challenges. Additionally, aging often comes with multiple chronic diseases, which increases the demand for healthcare services and burdens the existing healthcare system. Consequently, it is essential to identify sustainable strategies to promote care within this demographic [1, 8].
With technological advances, the Internet has emerged as an additional source of information for the public. Tools such as chatbots provide 24/7 access to information about dental and maxillofacial prostheses for geriatric patients and caregivers who need or use these prostheses. Artificial intelligence (AI) encompasses machines that perform various cognitive tasks associated with humans, including learning, problem-solving, and interacting with the environment. Advances in deep learning and other AI technologies contribute to the dynamic and continuous development of the AI field [8]. AI tools enable communication with generative AI and access to large language models (LLMs). Generative pre-trained transformers (GPT) are trained on large data sets and can generate responses to questions or engage in conversations initiated by humans [9]. During the creation of AI chatbots, raw data is used for pretraining. GPT is enhanced through supervised fine-tuning and reinforcement training. Sample data is processed based on the training data set to respond to a query, producing output through an algorithm. Due to variations in the training data set, different GPTs can yield different outputs.
Generative AI models such as Chat GPT-o1 (OpenAI, San Francisco, CA, USA) and DeepSeek-R1 (DeepSeek Inc., Hangzhou, China) comprehend language and content based on their respective AI data sets and generate new content using this information and reasoning. LLMs and AI chat interfaces are evolving with new data from prompts. ChatGPT-4 (OpenAI, San Francisco, CA, USA) is a general-purpose trained chatbot launched on March 14, 2023. It has been extensively tested in various environments for its conversational abilities and has demonstrated a high accuracy rate [10, 11]. ChatGPT-o1 was launched on September 12, 2024, featuring advanced reasoning skills that address complex queries in fields such as healthcare. This model aims to reduce hallucinations and provide more accurate outputs for complicated problems [12]. Dental GPT (OpenAI, San Francisco, CA, USA) is an interface specific to dentistry offered in June 2024 by OpenAI, which provides preliminary information to users and has the potential to support professional practices. DeepSeek-R1 was launched on January 20, 2025, as an alternative to OpenAI’s reasoning bots, claiming to deliver similar performance at a lower cost [13]. The development of generative AI chatbots and the introduction of healthcare-specific models highlight the necessity of scrutinizing the responses concerning their accuracy, quality, readability, understandability, and actionability. When evaluated, the dental implantology-related knowledge provided by Google Gemini, ChatGPT-3.5, and ChatGPT-4 revealed acceptable reliability and usefulness scores [14]. The reliability of ChatGPT-4’s responses was reported at 25. 6%, with repeatability ranging from substantial to moderate for professional questions about removable partial dentures and tooth-supported fixed partial dentures [15]. Several studies compared the information provided to patients by chatbots in orthodontics [16], orthognathic surgery [10], and dental trauma [11]. However, the accuracy, quality, readability, understandability, and actionability parameters of DeepSeek-R1, and ChatGPT-o1, which were recently launched with reasoning features, Dental GPT, which is specific to dentistry, and ChatGPT-4 have not been investigated and compared previously regarding patients’ queries about their prostheses. The accuracy of the responses in meeting patients’ needs is crucial, as incorrect information may lead to undesirable consequences and delays in referring to a physician [17].
This study aims to analyze and compare the answers of various chatbots, including generative, non-generative, and domain-specific models, to frequently asked questions from patients regarding their dental and maxillofacial prostheses, focusing on accuracy, quality, readability, understandability, and actionability. Therefore, the responses of DeepSeek-R1, Chat GPT-o1, ChatGPT-4, and Dental GPT to inquiries related to prosthodontics were evaluated. The null hypothesis posited that the responses generated by the four AI chatbots would not differ in accuracy, quality, readability, understandability, or actionability, nor would they adequately align with existing literature.
Methods
This study was conducted in silico without human or animal involvement or any patient-related information; therefore, Institutional Review Board approval was not required.
To determine the most submitted queries, the question “Which questions do people ask about dental and maxillofacial prostheses?” was posed to the DeepSeek-R1, ChatGPT-o1, ChatGPT-4, and Dental GPT chatbots. Two prosthodontists compared a total of 135 retrieved questions. Questions exhibiting content similarity (n = 52), repetition (n = 18), recommendations from only one chatbot (n = 12), and those relating to cost and insurance (n = 18) were eliminated. Ultimately, 35 questions in English language were selected and refined for presentation to the AI chatbots. These questions included subheadings addressing general inquiries, treatment and procedures, care and maintenance, and functional and aesthetic concerns regarding dental and maxillofacial prostheses (Table 1). Questions were also consulted to two other prosthodontists to validate the reflection of patient concerns in clinics.
Table 1.
List of questions asked to chatbots
| General Questions | |
|---|---|
| 1 | What types of dental prostheses are available for replacing missing teeth? |
| 2 | What is the difference between dentures, bridges, and implants? |
| 3 | Can I get a prosthesis if I have bone loss in my jaw? |
| 4 | Are dental implants safe for seniors? |
| 5 | Can I get implants if I have osteoporosis or diabetes? |
| 6 | Are there prostheses for facial reconstruction after surgery or injury? |
| 7 | What conditions require a maxillofacial prosthesis? |
| 8 | What are the different types of maxillofacial prostheses? |
| Treatment and Procedure | |
|---|---|
| 9 | Can dental prostheses be permanently attached or are they removable? |
| 10 | What materials are used in making dental prostheses? |
| 11 | Is getting a dental prosthesis procedure painful? |
| 12 | Will I need multiple appointments to get a dental prosthesis? |
| 13 | What are the stages for getting a dental prosthesis? |
| 14 | Are there common issues with dental prostheses that I should be aware of? |
| 15 | How long does it take to get used to a new dental prosthesis? |
| 16 | Are maxillofacial prostheses permanent or removable? |
| 17 | What are the stages for getting a maxillofacial prosthesis? |
| 18 | How does a prosthetic eye or nose stay in place? |
| Care & Maintenance | |
|---|---|
| 19 | How do I take care of my dental prosthesis? |
| 20 | How do I apply denture adhesive? |
| 21 | How long can I keep dentures in with adhesive? |
| 22 | Can I sleep with my dentures in? |
| 23 | When should I replace my dental prosthesis? |
| 24 | What should I do if my dental prosthesis breaks or gets damaged? |
| 25 | What should I do if my dentures become loose or uncomfortable? |
| 26 | How often do I need to see the dentist for check-ups after getting a prosthesis? |
| 27 | How do I clean my maxillofacial prosthesis? |
| 28 | Can a maxillofacial prosthesis wear out or need replacement? |
| Functional and Aesthetic Concerns | |
|---|---|
| 29 | Will I be able to speak normally with a dental prosthesis? |
| 30 | Can I eat normally with a dental prosthesis? |
| 31 | What foods should I avoid with dentures? |
| 32 | Can the dental prosthesis be customized to match my natural teeth? |
| 33 | Will I be able to speak normally with a maxillofacial prosthesis? |
| 34 | Can I eat normally with a maxillofacial prosthesis? |
| 35 | Can maxillofacial prostheses be custom-made to match natural features? |
The questions were posed to all chatbots by a volunteer who did not participate in the study in the same order and on the same day on February 9, 2025. A new chat window was opened for each question to prevent the influence of previous queries. Each response was saved in a Microsoft Word document (Microsoft Word, Version 16.0, Redmond, Washington: Microsoft Corporation). Each chatbot’s responses were color-coded before handling to researchers. Therefore, the researchers were blind to the name of the chatbot they were rating. At the end color codes corresponding to chatbots were revealed. An additional file shows a sample query in more detail [see Additional file 1].
Two prosthodontists rated all responses’ accuracy, quality, understandability, and actionability separately. A five-point Likert scale was used to rate the accuracy [16, 18]. Responses that were correct and accurate and comprised all relevant information on the subject were rated 5. The answer was rated as 1 if it was inconsistent with literature, with major mistakes, and potentially harmful to patient care [19].
The quality of the responses was evaluated using the Global Quality Scale (GQS), which featured a 5-point scoring system as outlined in previous studies [11, 20]. To assess readability, Flesch Reading Ease (FRE) and Flesch-Kincaid Grade Level (FKGL) scores were determined using a freely available online tool (https://goodcalculators.com/flesch-kincaid-calculator). The FRE and FKGL scales assign a numerical value to indicate the text’s difficulty level [11, 21]. Average sentence length (the mean number of words per sentence) and average word length (the mean number of syllables per word) are used to gauge readability. FRE and FKGL are calculated using the following formulas [11].
 |
 |
FRE values range from 0 to 100, with higher ratings signifying easier readability. FRE scores can be converted into estimated education levels using designated tables. FKGL indicates the minimum education level necessary for a manuscript to be easily read. Thus, a lower FKGL suggests that a lower education level is adequate for readers to comprehend the manuscript. Healthcare information documents are anticipated to achieve a rating of 8th -grade level or below [21].
The understandability and actionability of the manuscripts were evaluated using the Patient Education Materials Assessment Tool (PEMAT-P) [22]. This tool consists of 17 questions that assess the manuscript’s content for patient comprehension and an additional 7 questions if the manuscript provides instructions to guide the reader’s actions. Each item is rated separately as “Agree” if it meets the acceptable range of 80–100%, “Disagree,” or “Not Applicable” to derive the understandability and actionability scores [11].
Statistical analysis
All statistical analyses were conducted using SPSS for Windows, Version 21.0 (released 2012, IBM Corp., Armonk, NY, USA). Cohen’s kappa was calculated to evaluate the inter-rater agreement between the two researchers, yielding values of κ = 0.78 for the 5-point Likert scale, κ = 0.94 for the GQS, and κ = 0.88 for the PEMAT-P ratings, indicating a substantial and near-perfect level of agreement. The distribution of the obtained data was assessed using the Kolmogorov-Smirnov test. The Kruskal-Wallis and post hoc Mann-Whitney U tests were utilized to compare the parameters of accuracy, quality, readability, understandability, and actionability. One-way ANOVA was employed to analyze the FKGL parameter. Descriptive statistics were calculated to determine the median, minimum, maximum, mean, and standard deviations.
Results
The results are given in Table 2. Statistical analyses revealed significant differences between chatbots’ accuracy (p =.000) and FRE readability parameters (p =.034). Descriptive statistics showed that ChatGPT-o1 and Dental GPT achieved better accuracy scores than DeepSeek-R1 and ChatGPT-4 (p =.000).
Table 2.
Parameters and tools used to compare artificial intelligence chatbots and statistical analysis of scores
| Artificial Intelligence Chatbots | |||||
|---|---|---|---|---|---|
| Parameter- Tool |
DeepSeek-R1 | ChatGPT-o1 | ChatGPT-4 | Dental GPT | p |
|
Accuracy- § (5-point Likert) Median (Min-Max) |
5 (3–5)a | 5 (4–5)b | 5 (3–5)a | 5 (4–5)b | < 0.001 |
|
Quality- § GQS Median (Min-Max) |
5 (4–5) | 5 (4–5) | 5 (3–5) | 5 (4–5) | 0.191 |
|
Understandibility- § PEMAT-P Median (Min-Max) |
87.8 (63.6–92.3) |
88.4 (66.2–92.3) |
88.0 (52.2–100) |
90.8 (55.7–91.9) |
0.645 |
|
Actionability- § PEMAT-P Median (Min-Max) |
60 (20–70) | 60 (0–60) | 40 (0–80) | 40 (20–60) | 0.082 |
|
Readibility- § FRE Median (Min-Max) |
50.1 (17.8–71.6)a |
41.6 (14.9–63.7)ab |
45.2 (21.4–8.2)a |
37.7 (1.3–66.8)b |
< 0.05 |
|
Readibility- ŧ FKGL Mean ± SD |
9.4 ± 1.8 | 10.3 ± 1.8 | 10.5 ± 2.5 | 10.7 ± 2.7 | 0.095 |
§ Kruskal−Wallis Test. ŧ One−way ANOVA test. Lower−case letters ab display post hoc significant differences detected with Mann Whitney U test.
Abbreviations: GQS: Global Quality Score, PEMAT−P: Patient Education Materials Assessment Tool for Printed Materials, FRE: Flesch Reading Ease, FKGL: Flesch−Kincaid Grade Level, Min: Minimum, Max: Maximum, SD: Standard deviation
All the examined chatbots exhibited a similar median GQS value of 5, reflecting comparable excellent quality and flow. They were also deemed quite useful for patients (p =.191). The median understandability and actionability scores for chatbots ranged from 87.81 to 90.83 and 40 to 60, respectively, and the PEMAT-P scores of the chatbots showed no statistically significant differences for both parameters (p =.645, p =.082).
Figure 1 shows the distribution of responses regarding readability. DeepSeek-R1 had the highest median FRE score reflecting readability, while Dental GPT exhibited the lowest performance. Post hoc analysis revealed that Dental GPT performed significantly worse in readability than the other models (p =.034). The FKGL values of all chatbots ranged between 9.42 and 10.70, displaying no significant difference. The readers’ estimated educational levels were at least the eighth and ninth grades.
Fig. 1.
Distribution of responses for readability
Discussion
In this study, the accuracy, quality, readability, understandability, and actionability of the responses regarding dental and maxillofacial prostheses provided by DeepSeek-R1, ChatGPT-o1, ChatGPT-4, and Dental GPT were statistically tested. The results indicated that the quality, understandability, and actionability parameters of the null hypothesis was accepted. In contrast, the accuracy and readability aspects were rejected at the specified significance level. These results are significant as they demonstrate the importance of selecting a chatbot to access accurate and easily readable patient information.
In the literature, the performances of general chatbots such as ChatGPT, Gemini, and Copilot, as well as healthcare-specific bots such as Openevidence and Medisearch, have been evaluated across various healthcare fields, including orthodontics [16], orthognathic surgery [10], dental trauma [11], tooth extraction [18], oral and maxillofacial radiology [21] and dental prostheses [15]. This study is among the first to evaluate DeepSeek-R1 and ChatGPT-o1 with generative AI features, including ChatGPT-4 for general use and Dental GPT chatbots specifically designed for prosthodontics patients.
Accuracy defines how close a given set of measurements is to the true values. In our study, the compliance of the answers provided by the chatbots with the current literature and guidelines was evaluated using a 5-point Likert scale. Although the median accuracy rating of all chatbots was 5, the accuracy ratings varied statistically, with Dental GPT and ChatGPT-o1 achieving significantly higher scores. The high accuracy of Dental GPT, which was developed specifically for the dental field to benefit patients, highlights the significance of the chatbot’s dataset. These results are consistent with previous reports indicating that chatbots trained with domain-specific datasets outperform general-purpose LLMs such as ChatGPT-4 [23]. It was noted that ChatGPT-4 has limited ability to answer professional questions regarding fixed and removable prostheses, and users should be aware of these limitations [15]. The accuracy of ChatGPT-4 in the present study was higher when compared to previous studies, as the questions addressed more basic and straightforward concepts that do not require extensive reasoning. However, its shortcomings in providing evidence-based information should not be overlooked. In a study assessing the accuracy, reliability, quality, and readability of information given to patients by ChatGPT-3.5, ChatGPT-4, Gemini, and Copilot AI concerning orthodontic clear aligners, it was noted that all chatbot models generally provided accurate, moderately reliable responses of moderate to good quality, demonstrating significant potential as patient information tools [16].
Patients with limited mobility may not need to see a doctor for issues resolvable through accurate information. This can reduce healthcare service demands and alleviate pressure on the system. While chatbot responses for patient education are often accurate, potential negative outcomes due to incorrect answers must be considered. For example, DeepSeek-R1 has provided inaccurate responses to predictable questions, such as the coloring of maxillofacial prostheses. Both healthcare professionals and patients should be cautious with chatbot information, ensuring verification and improved testing mechanisms are in place.
Readability is the extent to which a written text is easy for a reader to understand. Factors like the reader’s intelligence, education, environment, interests, purpose, and the text’s ideas, vocabulary, style, and format impact readability. In the current study, the median FRE score of Dental GPT was 37.7, significantly lower than that of other chatbots, indicating the worst readability. FKGL scores of all chatbots ranged from 9.42 to 10.70, suggesting a fairly difficult reading level. Chatbots used for patient information and guidance should ideally have high FRE and low FKGL scores to benefit users. Although DeepSeek-R1 and ChatGPT-4 appear statistically easier to read, all artificial intelligence chatbots were found to have an estimated reading grade level equivalent to 8th and 9th grade. This level exceeds the education required for health-related documents [21]. In conclusion, the level of public education is crucial for patients to be able to easily read and benefit from the answers provided by chatbots. Cognitive functions may decline with advancing age, and caregivers of elderly patients must also possess a high level of education to guide patients effectively. The readability of all chatbots, particularly Dental GPT, should be enhanced, as it provides the most accurate and reliable answers.
In this study, the differences in quality, comprehensibility and actionability scores were not statistically significant. ChatGPT-4, Medisearch, and Openevidence were reported to provide good-quality responses to patient information about orthognathic surgery [10]. In another study evaluating ChatGPT-3.5, ChatGPT-4, and Gemini, the responses of ChatGPT-4 were reported to be higher than those of the others [11]. A comparison is inappropriate since only ChatGPT-4 was common in the current study. However, the answers to the questions were of very high quality, flowed well, and were very helpful to the patients.
The understandability values of the texts, evaluated as percentages using the PEMAT-P scale, ranged from 87.81 to 90.8%, while the actionability values varied between 40% and 60%. All understandability values exceeded the threshold of 70%, consistent with the previous study on traumatic dental injuries [11]. These results suggest that the texts were quite understandable and could be 40–60% effective in guiding patients to act. The relatively low actionability value may be attributed to the nature of the prosthodontics domain. For both new prostheses and the maintenance of existing ones, patients with prosthetic needs must consult their dentist. Typically, patients have their current prosthesis fitted in their mouth and receive instructions on better maintaining oral hygiene. When geriatric patients struggle to meet with the dentist and remember the provided recommendations, chatbot responses can alleviate their curiosity, guide them in caring for their dentures, and encourage them to see the dentist when necessary. All chatbots responded accurately to inquiries about prosthesis care and directed patients to seek dental assistance when needed. The strengths, similarities, and weaknesses of the information provided by LLMs were evaluated for geriatric patients regarding dental and maxillofacial prostheses. Even when the answers included the appropriate information, all chatbots typically supplied more than necessary, which might confuse patients. Researchers’ impressions of chatbots are as follows: DeepSeek-R1 generally answered the questions correctly but could not accurately determine the order of coloring during the construction of the maxillofacial prosthesis. It also did not mention the orbital prosthesis while informing the patient about facial restoration. ChatGPT-o1 provided long, detailed, and informative answers to the questions, stating that it was not a healthcare professional and directing the patient to consult a prosthodontist or a medical team. It did not mention the option of cranial implants concerning the reconstruction of facial prostheses. ChatGPT-4 offered short, easy- to- understand answers about prosthesis renewal and maintenance. Apart from the original responses, Dental GPT asked and answered additional detailed questions the patient might be curious about. It also provided more detailed answers than other chatbots regarding the wear and renewal of the maxillofacial prosthesis.
Artificial intelligence tools can assist prosthodontists with dental charting, tooth color selection, restoration design, and predicting changes in facial contours. However, they are still under development [24]. These tools may enhance treatment planning and decision-making while improving patient communication through chatbots. They can also save physicians’ time by automating routine tasks, such as customer service and scheduling, allowing a focus on more creative or strategic work. This study follows the launch of DeepSeek-R1 and may be among the first to use this AI in prosthodontics. Launched at a lower cost, it claims similar performance to ChatGPT4 and has shown comparable accuracy with superior readability among chatbots. It offers an alternative to existing OpenAI tools. All AI technologies are evolving rapidly, and their accuracy may improve with larger datasets. The study results were performed with the latest tools at the time of the study. However, ChatGPT 4.5 was launched after the data analysis, and the results do not represent this tool. Another limitation of the study was that it was conducted only in English. Chatbots’ answers to questions asked in different languages may differ in performance. Until DeepSeek was released, English was the dominant language in LLM training. DeepSeek was trained primarily in Chinese and English, and the answers generated by combining the two languages during reasoning are thought to be more comprehensive than those generated by selecting a single language.
Although they were among the most frequently asked topics, questions regarding cost and insurance were excluded from the study as the reference comparison data would vary significantly depending on the countries and their social security systems.
Conclusions
Although all chatbots provided access to accurate information, domain-specific GPT offered more precise details than general-purpose chatbots. However, the readability of Dental GPT was significantly lower. There is a need to enhance the readability of the responses. The chatbots examined were very useful for patients, exhibiting excellent quality and flow.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgements
Not applicable.
Abbreviations
- AI
Artificial intelligence
- LLM
Large language model
- GPT
Generative pre-trained transformers
- GQS
Global Quality Scale
- FRE
Flesch Reading Ease
- FKGL
Flesch-Kincaid Grade Level
Author contributions
T.Ö. conceptualized the study. Methodology, formal analysis and investigation, writing - original draft preparation, review and editing was performed by T.Ö. and B.Ö. All authors reviewed the manuscript.
Funding
The authors did not receive any funding support from any organization for the submitted work.
Data availability
The data supporting this study’s findings are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
Not applicable. As this in silico study does not involve any material or data related with human participants or animals, ethical approval is not required and applicable.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Ma B, Yang J, Wong FKY, Wong A, Ma T, Meng J, et al. Artificial intelligence in elderly healthcare: A scoping review. Ageing Res Rev. 2023;83:101808. 10.1016/j.arr.2022.101808. [DOI] [PubMed] [Google Scholar]
- 2.(2022) World Health Organization (2022) Global oral health status report: towards universal health coverage for oral health by 2030. WHO Team. https://www.who.int/publications/i/item/9789240061484 Accessed 19 February 2025.
- 3.Scannapieco FA. Poor oral health in the etiology and prevention of aspiration pneumonia. Clin Geriatr Med. 2023;39:257–71. 10.1016/j.cger.2023.01.010. [DOI] [PubMed] [Google Scholar]
- 4.Singh YP, Saini M, Rehman SU. The effect of complete denture rehabilitation on nutritional status of geriatric patients. J Nat Sci Med. 2021;4:274–80. 10.4103/jnsm.jnsm_1_21. [Google Scholar]
- 5.Kalyoncuoğlu ÜT, Saraç Atagün Ö. Evaluation of the hygiene habits and awareness of the removable denture wearer geriatric population: A cross-sectional survey study. Northwest Med J. 2023;3:156–64. 10.54307/NWMJ.2023.94830. [Google Scholar]
- 6.Bisht S, Mao Y, Easwaran H. Epigenetic dynamics of aging and cancer development: current concepts from studies mapping aging and cancer epigenomes. Curr Opin Oncol. 2024;36:82–92. 10.1097/CCO.0000000000001020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kemp M, Wallace F, Henry E. (2012) Public Perceptions of Choice in UK Dental Care. British Dental Association. https://www.researchgate.net/publication/376394469_Public_Perceptions_of_Choice_in_UK_Dental_Care Accessed 19 February 2025.
- 8.Koç M. Artificial intelligence in geriatrics. Turkish J Geriatr. 2023;26:352–60. 10.29400/tjgeri.2023.362. [Google Scholar]
- 9.Clark M, Bailey S. (2024) CADTH Horizon Scan Chatbots in Health Care: Connecting Patients to Information. NIH Publication. https://www.ncbi.nlm.nih.gov/books/NBK602381/pdf/Bookshelf_NBK602381.pdf Accessed 11 March 2025. [PubMed]
- 10.Yurdakurban E, Topsakal KG, Duran GS. A comparative analysis of AI-based chatbots: assessing data quality in orthognathic surgery related patient information. J Stomatol Oral Maxillofac Surg. 2024;125. 10.1016/j.jormas.2023.101757. [DOI] [PubMed]
- 11.Guven Y, Ozdemir OT, Kavan MY. (2024) Performance of artificial intelligence chatbots in responding to patient queries related to traumatic dental injuries: A comparative study. Dental Traumatology. 10.1111/edt.13020 [DOI] [PubMed]
- 12.Temsah M-H, Jamal A, Alhasan K, Temsah A, Malki K. OpenAI o1-Preview vs. ChatGPT in healthcare: A new frontier in medical AI reasoning. Cureus. 2024. 10.7759/cureus.70640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.DeepSeek-AI, Guo D, Yang D, Zhang H, Song J, Zhang R et al. (2025) DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. 10.48550/arXiv.2501.12948
- 14.Taymour N, Fouda SM, Abdelrahaman HH, Hassan MG. Performance of the ChatGPT-3.5, ChatGPT-4, and Google gemini large Language models in responding to dental implantology inquiries. J Prosthet Dent. 2025. 10.1016/j.prosdent.2024.12.016. [DOI] [PubMed] [Google Scholar]
- 15.Freire Y, Santamaría Laorden A, Orejas Pérez J, Gomez Sanchez M, Diaz-Flores Garcia V, et al. ChatGPT performance in prosthodontics: assessment of accuracy and repeatability in answer generation. J Prosthet Dent. 2024;131. 10.1016/j.prosdent.2024.01.018.:659.e1-659.e6. [DOI] [PubMed]
- 16.Dursun D, Bilici Geçer R. Can artificial intelligence models serve as patient information consultants in orthodontics? BMC Med Inf Decis Mak. 2024. 10.1186/s12911-024-02619-8. 24:. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cuevas-Nunez M, Silberberg VIA, Arregui M, Jham B, Baallester-Victoria R, Koptseva I, et al. Diagnostic performance of ChatGPT-4.0 in histopathological description analysis of oral and maxillofacial lesions: a comparative study with pathologists. Oral Surg Oral Med Oral Pathol Oral Radiol. 2025;139:453–61. 10.1016/j.oooo.2024.11.087. [DOI] [PubMed] [Google Scholar]
- 18.Jacobs T, Shaari A, Gazonas CB, Ziccardi VB. Is ChatGPT an accurate and readable patient aid for third molar extractions?? J Oral Maxillofac Surg. 2024;82:1239–45. 10.1016/j.joms.2024.06.177. [DOI] [PubMed] [Google Scholar]
- 19.Victorian Government Department of Human Services. (2002) Oral Health for Older People A Practical Guide for Aged Care Services. Rural and Regional Health and Aged Care Services Division. https://www.dhsv.org.au/__data/assets/pdf_file/0003/151869/oral-health-for-older-people.pdf Accessed 19 February 2025.
- 20.Bernard A, Langille M, Hughes S, Rose C, Leddin D, van Veldhuyzen S. A systematic review of patient inflammatory bowel disease information resources on the world wide web. Am J Gastroenterol. 2007;102:2070–7. 10.1111/j.1572-0241.2007.01325.x. [DOI] [PubMed] [Google Scholar]
- 21.Helvacioglu-Yigit D, Demirturk H, Ali K, Tamimi D, Koenig L, Almashraqi A. Evaluating artificial intelligence chatbots for patient education in oral and maxillofacial radiology. Oral Surg Oral Med Oral Pathol Oral Radiol. 2025. 10.1016/j.oooo.2025.01.001. [DOI] [PubMed] [Google Scholar]
- 22.Shoemaker SJ, Wolf MS, Brach C. Development of the patient education materials assessment tool (PEMAT): A new measure of understandability and actionability for print and audiovisual patient information. Patient Educ Couns. 2014;96:395–403. 10.1016/j.pec.2014.05.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lehman E, Hernandez E, Mahajan D, Wulff J, Smith M, Ziegler Z, et al. Do we still need clinical Language models?? Proc Mach Learn Res. 2023;209:578–97. 10.48550/arXiv.2302.08091. [Google Scholar]
- 24.Revilla-León M, Gómez-Polo M, Vyas S, Barmak A, Gallucci G, Att W, et al. Artificial intelligence models for tooth-supported fixed and removable prosthodontics: A systematic review. J Prosthet Dent. 2023;129:276–92. 10.1016/j.prosdent.2021.06.001. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data supporting this study’s findings are available from the corresponding author upon reasonable request.