Performance of three artificial intelligence (AI)-based large language models in standardized testing; implications for AI-assisted dental education.

INTRODUCTION: The emerging rise in novel computer technologies and automated data analytics has the potential to change the course of dental education. In line with our long-term goal of harnessing the power of AI to augment didactic teaching, the objective of this study was to quantify and compare the accuracy of responses provided by ChatGPT (GPT-4 and GPT-3.5) and Google Gemini, the three primary large language models (LLMs), to human graduate students (control group) to the annual in-service examination questions posed by the American Academy of Periodontology (AAP). METHODS: Under a comparative cross-sectional study design, a corpus of 1312 questions from the annual in-service examination of AAP administered between 2020 and 2023 were presented to the LLMs. Their responses were analyzed using chi-square tests, and the performance was juxtaposed to the scores of periodontal residents from corresponding years, as the human control group. Additionally, two sub-analyses were performed: one on the performance of the LLMs on each section of the exam; and in answering the most difficult questions. RESULTS: ChatGPT-4 (total average: 79.57%) outperformed all human control groups as well as GPT-3.5 and Google Gemini in all exam years (p < .001). This chatbot showed an accuracy range between 78.80% and 80.98% across the various exam years. Gemini consistently recorded superior performance with scores of 70.65% (p = .01), 73.29% (p = .02), 75.73% (p < .01), and 72.18% (p = .0008) for the exams from 2020 to 2023 compared to ChatGPT-3.5, which achieved 62.5%, 68.24%, 69.83%, and 59.27% respectively. Google Gemini (72.86%) surpassed the average scores achieved by first- (63.48% ± 31.67) and second-year residents (66.25% ± 31.61) when all exam years combined. However, it could not surpass that of third-year residents (69.06% ± 30.45). CONCLUSIONS: Within the confines of this analysis, ChatGPT-4 exhibited a robust capability in answering AAP in-service exam questions in terms of accuracy and reliability while Gemini and ChatGPT-3.5 showed a weaker performance. These findings underscore the potential of deploying LLMs as an educational tool in periodontics and oral implantology domains. However, the current limitations of these models such as inability to effectively process image-based inquiries, the propensity for generating inconsistent responses to the same prompts, and achieving high (80% by GPT-4) but not absolute accuracy rates should be considered. An objective comparison of their capability versus their capacity is required to further develop this field of study.

Periodontists and stem cell-based therapy for alveolar bone regeneration: A national survey.

BACKGROUND: Stem cell-based therapy for bone regeneration has received attention in medical settings but has not yet been used in clinical practice for treating alveolar bone defects. The objectives of this study were to explore whether periodontists had heard about this approach, and if so how, how interested they were to learn about it, which attitudes and behavioral intentions they had related to using stem cell-based grafting, and what they would like to know before using this approach. METHODS: Anonymous survey data were collected from 481 members of the American Academy of Periodontology (response rate: 19.41%). RESULTS: Responses showed 35.3% had heard about stem cell-based therapy, mostly from publications (9.6%) and meetings (8.3%); 76.1% wanted to learn about it through in-person continuing education (CE) courses, 68.6% in online CE courses, and 57.1% from manuals; 73% considered this approach promising; and 54.9% preferred it to traditional approaches. It was important to them that it would result in more bone volume (93%), better bone quality (90.4%), and accelerated healing (83.2%). Also, 60.1% considered it likely/very likely that they would adopt this approach, 54% that patients would prefer it, and 62.1% that it would benefit their practice. When asked what they would like to know about this approach, information about short- and long-term outcomes, cost, and logistical considerations were most frequently named. CONCLUSIONS: These findings provide the basis to develop educational interventions for periodontists about this novel approach and inform future research activities aimed to translate this approach to clinical practice.

Capnography monitoring in procedural intravenous sedation: a systematic review and meta-analysis.

OBJECTIVES: Currently, procedural sedation in the clinical setting relies heavily on the use of pulse oximetry to monitor hypoxemia. Different studies suggest that incidence of hypoxemia and incidence of arterial oxygen desaturation are reduced by early intervention via capnography monitoring. The aim of this article was to discuss the importance of implementing capnography monitoring during procedural sedations performed in a dental setting and determine whether additional capnographic monitoring reduces the incidence of arterial oxygen desaturation and the overall complications rate. MATERIALS AND METHODS: Two independent reviewers conducted electronic (PubMed and EMBASE) and manual searches up to February 2020. Randomized clinical trials (RCTs), including both patients under procedural sedation monitored by capnography and oximetry, and reporting the incidence of hypoxemia or episodes of oxygen desaturation were included. Risk ratio was used to compare the outcomes (i.e., the incidence of hypoxemia, the episodes of oxygen desaturation, the detection of apnea, the reduction of events of bradycardia, and hypotension) between patients monitored by capnography and standard approach. RESULTS: Fourteen randomized clinical trials fulfilling the inclusion criteria were selected. The analysis revealed that capnography monitoring group showed the lower incidence of hypoxemia (RR 0.76, 95%CI 0.70 to 0.83, p < 0.001) and the episodes of oxygen desaturation (RR 0.79, 95%CI 0.71 to 0.87, p < 0.001) compared with the oximetry monitoring group. Apnea was detected in capnography monitoring earlier than standard monitoring (RR 2.60, 95%CI 2.30 to 2.93, p < 0.001). No significant difference was found between capnography and standard monitoring groups in terms of reduction of events of bradycardia (RR 1.17, 95%CI 0.91 to 1.50, p = 0.225) and hypotension (RR 0.96, 95%CI 0.76 to 1.21, p = 0.746). CONCLUSION: Capnography monitoring reduced incidence of hypoxemia during procedural sedations. Within the limitations of this review, we suggest that the application of capnography during procedural sedation would decrease the frequency of oxygen desaturation events and incidence of hypoxemia. CLINICAL RELEVANCE: Training and instructing dental providers on using capnography monitoring would help in reducing adverse events during intravenous sedation.

Breaking with Tradition: A Scoping Meta-Analysis Analyzing the Effects of Student-Centered Learning and Computer-Aided Instruction on Student Performance in Anatomy.

While prior meta-analyses in anatomy education have explored the effects of laboratory pedagogies and histology media on learner performance, the effects of student-centered learning (SCL) and computer-aided instruction (CAI) have not been broadly evaluated. This research sought to answer the question, "How effective are student-centered pedagogies and CAI at increasing student knowledge gains in anatomy compared to traditional didactic approaches?" Relevant studies published within the past 51 years were searched using five databases. Predetermined eligibility criteria were applied to the screening of titles and abstracts to discern their appropriateness for study inclusion. A summary effect size was estimated to determine the effects of SCL and CAI on anatomy performance outcomes. A moderator analysis of study features was also performed. Of the 3,035 records screened, 327 underwent full-text review. Seven studies, which comprised 1,564 participants, were included in the SCL analysis. An additional 19 studies analyzed the effects of CAI in the context of 2,570 participants. Upon comparing SCL to traditional instruction, a small positive effect on learner performance was detected (standardized mean difference (SMD = 0.24; [CI = 0.07, 0.42]; P = 0.006). Likewise, students with CAI exposure moderately outscored those with limited or no access to CAI (SMD = 0.59; [CI = 0.20, 0.98]; P = 0.003). Further analysis of CAI studies identified effects (P ≤ 0.001) for learner population, publication period, interventional approach, and intervention frequency. Overall, learners exposed to SCL and supplemental CAI outperformed their more classically-trained peers as evidenced by increases in short-term knowledge gains. Anat Sci Educ. © 2018 American Association of Anatomists.

Elaboration of tubules with active hedgehog drives parenchymal fibrogenesis in gestational alloimmune liver disease.

Gestational alloimmune liver disease (GALD) produces severe neonatal liver disease that is notable for paucity of hepatocytes, large numbers of parenchymal tubules, and extensive fibrosis. Liver specimens from 19 GALD cases were studied in comparison with 14 infants without liver disease (normal newborn liver; NNL) to better understand the pathophysiology that would produce this characteristic histopathology. GALD liver parenchyma contained large numbers of tubules comprising epithelium expressing KRT7/19, EPCAM, and SOX9, suggesting biliary progenitor status. Quantitative morphometry demonstrated that in GALD, the area density of KRT19+ tubules was 16.4 ± 6.2 versus 2.0 ± 2.6 area% in NNL (P < .0001). Functional hepatocyte mass was markedly reduced in GALD, 16.3 ± 6.2 versus 61.9 ± 11.0 area% of CPS1+ cells in NNL (P < .0001). A strong inverse correlation was established between CPS1+ area density and KRT19+ area density (r(2) = 0.66, P < .0001). Tubules showed active hedgehog signaling as determined by SHH and nuclear GLI2 expression and expressed the profibrogenic cytokine SPP1. SPP1 protein content and SPP1 expression were greater in GALD than NNL (15- and 13-fold respectively; P = .002). GALD liver contained large numbers of activated myofibroblasts and showed greater than 10-fold more fibrosis than NNL. The extent of fibrosis correlated with the area density of KRT19+ tubules (r(2) = 0.387, P = .001). The data support a pathogenic model in which immune injury to fetal hepatocytes provides a stimulus for expansion of parenchymal tubules, which, by way of Hh activation, produce fibrogenic signals leading to vibrant fibrosis.