{"@context":"https://schema.org","@type":"ScholarlyArticle","headline":"Emergency Management of Permanent Tooth Avulsion: Comparative Performance of Dental Students and Artificial Intelligence Using a Multimodal Clinical Vignette Across Bloom's Taxonomy Domains.","author":[{"@type":"Person","name":"Sanaa N Al-Haj Ali"}],"datePublished":"2025","isPartOf":{"@type":"Periodical","name":"International endodontic journal"},"identifier":"https://doi.org/10.1111/iej.70060","description":"ABSTRACT\n \n Aim\n \n To pioneer the first comprehensive benchmarking of dental students against 22 multimodal Artificial Intelligence (\n AI\n ) models spanning six major architectures (Anthropic,\n DeepSeek\n , Google Gemini,\n OpenAI\n , Meta Llama, Mistral/Qwen) in emergency permanent tooth avulsion management, quantifying performance differences across Bloom's Taxonomy higher‐order cognitive domains (Apply, Analyse, Evaluate) against International Association of Dental Traumatology (\n IADT\n ) guidelines to optimise evidence‐based educational integration.\n \n \n \n Methodology\n \n This cross‐sectional study compared 35 fifth‐year dental students with 22 multimodal\n AI\n models grouped by producer (Anthropic,\n DeepSeek\n , Google Gemini,\n OpenAI\n , Meta, Mistral/Qwen) using a validated clinical vignette featuring a 14‐year‐old with three avulsed mature permanent teeth (#11, 12, 22) with clinical photograph. Performance was assessed through four questions (1 Apply, 2 Analyse, 1 Evaluate) via an\n IADT\n ‐aligned rubric validated against paediatric dentistry standards. Responses were independently scored by two blinded evaluators. Subsequently, scores were compared using\n ANOVA\n with post hoc Tukey\n HSD\n , Kruskal‐Wallis, and chi‐square tests (\n SPSS\n v20,\n α\n = 0.05).\n \n \n \n Results\n \n Overall performance analysis revealed significant group differences (\n p\n = 0.012, partial\n η\n \n 2\n \n = 0.27). Only students achieved high‐acceptability A‐level performance (17.1%,\n χ\n 2\n = 39.27,\n p\n = 0.003), despite top\n AI\n models (Claude‐Sonnet‐3.7‐Reasoning,\n DeepSeek\n ‐\n R1\n : 7.0/10) marginally exceeding student mean scores (6.5/10). Performance varied among\n AI\n systems, with students significantly outperforming Meta Llama models (mean difference = 3.50,\n p\n = 0.014), which showed exclusively not acceptable (D‐level) performance. Students dominated Bloom's Analyse level (visual diagnosis: 100% tooth identification vs. 77.3%\n AI\n failure; age‐specific management differentiation,\n p\n < 0.001), while\n AI\n excelled in Apply/Emergency steps (medical management: 81.8% perfect scores) and Evaluate/prosthetic contraindication reasoning (Anthropic: 0.92 ± 0.20 vs. students: 0.41 ± 0.33,\n p\n = 0.005). Critical\n AI\n deficiencies included tooth misidentification (77% of models), failure to mention follow‐up schedules for immature replanted teeth (100%), and inappropriate intervention recommendations (40.9% suggested unsuitable options like resin‐bonded bridges). For soft tissue management, only 23% of\n AI\n models addressed the need to suture gingival lacerations.\n \n \n \n Conclusions\n \n AI\n 's fundamental limitations in visual diagnosis,\n IADT\n misinterpretations, and inadequate clinical translation mandate strict positioning as a supplemental protocol accelerator—not an independent diagnostic tool. Educational integration requires: (1) Careful model selection and supervision for any visual triage applications; (2) Mandatory\n IADT\n verification; and (3) Preservation of human clinical reasoning as the irreplaceable cornerstone of permanent tooth avulsion management.\n \n ","url":"https://innscience.ca/es/community/literature/2636d17f-9f45-4466-957d-62e35b4e6b1b"}