Prita Abhay Dhaimade

Multimodal large language models (MLLMs) are increasingly applied to image-based clinical reasoning, yet their diagnostic reliability in periodontal image interpretation, and the underlying source of their errors, remain poorly characterized. This study evaluated six architecturally distinct MLLMs (Claude Sonnet 4.5, GPT-5.0, Gemini 2.5, GLM-4.6, Sonar, and Grok 4.1) using 50 image-based multiple-choice questions drawn from the American Academy of Periodontology In-Service Examination, spanning clinical photographs, histopathology, radiographs, cardiac rhythm strips, and anatomical illustrations. A sequential two-phase experimental design was used: in Phase 1, each model independently described each image, selected an answer, and provided a supporting citation; in Phase 2, applied only to questions answered incorrectly, models were given an expert-validated visual description and asked to re-answer, allowing diagnostic improvement through visual correction to be measured directly. Expert ground truth for image content was established by a board-certified periodontist and independently validated by a second board-certified periodontist. Model outputs were classified using a dual-process error taxonomy adapted from Norman’s model of diagnostic reasoning, distinguishing perceptual errors, arising from inaccurate visual feature extraction, from cognitive errors, arising from flawed reasoning despite accurate perception, with cognitive errors further subdivided into correctable and persistent subtypes, and additional categories capturing compound perceptual-cognitive failures and compensatory reasoning that overcame inaccurate perception. Diagnostic accuracy and error type distribution varied significantly across models and image modality. Correcting inaccurate visual descriptions in Phase 2 improved diagnostic accuracy for a subset of previously incorrect responses, indicating that a meaningful share of errors originated at the level of visual perception rather than clinical reasoning; conversely, a distinct subset of errors persisted despite accurate corrected visual input, indicating reasoning-level failures independent of perceptual accuracy. Some models also reached correct answers despite generating inaccurate image descriptions, reflecting compensatory reasoning resilient to perceptual error. These findings show that aggregate accuracy scores conflate mechanistically distinct failure modes, and that perceptual and cognitive errors carry different implications for how MLLMs might be safely deployed or improved for diagnostic image interpretation. The expert-guided visual correction framework introduced here provides a generalizable, mechanism-based approach to benchmarking multimodal AI diagnostic performance that extends beyond periodontics to other visually driven diagnostic domains in medicine. As MLLMs become increasingly accessible to clinicians, residents, and dental educators, distinguishing perceptual from cognitive failure is essential for guiding responsible clinical use, targeting model refinement, and informing AI-augmented dental education and competency assessment.

Large language models (LLMs) have demonstrated rapid advancements in natural language understanding and generation, prompting their integration into biomedical research, clinical practice, and professional education. However, systematic evaluation of LLMs in specialty-specific domains such as dentistry and periodontology remains limited, particularly regarding multidimensional performance metrics. This study conducted a comprehensive assessment of commercially available LLMs — GPT-4.0, GPT-5.0, and Claude Sonnet 4.0 — on the American Academy of Periodontology In-Service Examination, focusing on response accuracy, self-assessed confidence calibration, citation validity, and hallucination prevalence. Models were evaluated on the 2024 AAP In-Service Examination (331 questions) using two formats: Full Test (all questions at once) and Individual Question (one at a time). Prompts were standardized; models selected answers, and GPT-5.0 and Claude Sonnet 4.0 also provided confidence ratings and citations. Citation validity was assessed using a human-in-the-loop protocol with expert review. Statistical analyses included chi-square, McNemar’s, and logistic regression to assess accuracy, question fatigue, confidence calibration, and citation reliability. LLMs achieved high overall accuracy (78–87%), with the Individual Question format consistently yielding higher scores than Full Test, though differences were not statistically significant. Accuracy was highest in fact-dense domains (biochemistry, physiology, microbiology) and lowest in integrative domains (diagnosis, therapy). Significant question fatigue was observed in GPT-5.0 Full Test mode (OR = 0.997, p = 0.035) but not in Individual Question mode. Confidence scores predicted accuracy, with the strongest calibration in Individual Question mode. Citation analysis revealed frequent hallucinations, mostly critically erroneous, and citation validity was independent of answer accuracy. LLMs show promise as adjunctive tools for periodontal education, but their outputs, especially for complex reasoning and citations require rigorous human review to ensure accuracy and safety.

Background: Large language models (LLMs) have demonstrated rapid advancements in natural language understanding and generation, prompting their integration into biomedical research, clinical practice, and professional education. However, systematic evaluation of LLMs in specialty-specific domains such as dentistry and periodontology remain limited, particularly regarding multidimensional performance metrics. Objective: To conduct a comprehensive, multidimensional assessment of commercially available LLMs: GPT-4.0, GPT-5.0, and Claude SONNET 4.0 on the American Academy of Periodontology in-service examination, focusing on response accuracy, self-assessed confidence calibration, citation validity, and hallucination prevalence. Methods: Models were evaluated on the 2024 AAP In-Service Examination (331 questions) using two formats: Full Test (all questions at once) and Individual Question (one at a time). Prompts were standardized; models selected answers, and for GPT-5.0 and Claude SONNET 4.0, also provided confidence ratings and citations. Citation validity was assessed using a human-in-the-loop protocol with expert review. Statistical analyses included chi-square, McNemar’s, and logistic regression to assess accuracy, question fatigue, confidence calibration, and citation reliability. Results: LLMs achieved high overall accuracy (78–87%), with the Individual Question format consistently yielding higher scores than Full Test, though differences were not statistically significant. Accuracy was highest in fact-dense domains (biochemistry, physiology, microbiology) and lowest in integrative domains (diagnosis, therapy). Significant question fatigue was observed in GPT-5.0 Full Test mode (OR = 0.997, p = 0.035), but not in Individual Question mode. Confidence scores predicted accuracy, with the strongest calibration in Individual Question mode. Citation analysis revealed frequent hallucinations, mostly critically erroneous, and citation validity was independent of answer accuracy. Conclusions: LLMs can answer a broad spectrum of periodontal specialty questions, but their reliability varies with context and information presentation. While promising as adjunctive tools, their outputs— especially for complex reasoning and citations—require rigorous human review in educational and research settings to ensure accuracy and safety. Author Summary: Artificial intelligence chatbots are rapidly entering medical education, yet we lack comprehensive understanding of their reliability when students depend on them for learning. We developed a multidimensional evaluation framework to systematically assess AI performance beyond simple accuracy, examining how these systems behave across different medical topics, question types, and presentation formats. Using 331 real dental examination questions, we tested three major AI systems, analyzing not only correctness but also confidence calibration - whether AI confidence levels match actual accuracy - and implementing human-in-the-loop verification to check if cited sources actually exist. Our findings highlight critical vulnerabilities in current AI systems. Most alarmingly, these chatbots fabricated nearly half of their citations while maintaining unwavering confidence in both correct and incorrect responses. This combination of overconfidence and misinformation means students cannot distinguish reliable from unreliable AI responses. Additionally, we documented progressive performance decline during sequential questioning, similar to human cognitive fatigue. While we know AI systems generate rather than retrieve information, our research demonstrates the real-world consequences of this limitation. As artificial intelligence integrates into education, healthcare diagnostics, and insurance decisions, these findings underscore the urgent need for better evaluation frameworks and user education about AI limitations.

Introduction: Burnout syndrome refers to a combination of physical fatigue and emotional exhaustion, which, in turn, affects the working efficiency of a person. In India, factors such as extensive working hours, poor facilities, and physical and emotional abuse of doctors by patients and seniors lead to the high prevalence of occupational burnout among medical practitioners. Materials and Methods: The sample consisted of 300 resident doctors working in public sector hospitals across Mumbai. The “Copenhagen Burnout Inventory” questionnaire was utilized to assess the prevalence of burnout. Questionnaires were made available personally or electronically. Burnout was recorded on three parameters, personal burnout, work-related burnout, and client-related burnout. Results: The average working hours recorded was 88 h/week. About 56.66% (n = 170) showed scores that indicate burnout. About 66.67% of respondents showed personal burnout, 57.14% had work-related burnout, and 16.67 had client-related burnout. Conclusion: The high prevalence of burnout syndrome among resident doctors in public sector hospitals is alarming as it not only takes a toll on the physical and mental health of the medical practitioners but also reduces their working efficiency and motivation. Stress management strategies should be propagated in hospitals to encourage work and personal life balance.

The purpose of this study was to conduct a comparative assessment of the various cephalometric and auditory parameters between patients with Down syndrome (DS) and healthy controls. The cephalometric and auditory parameters were divided among 50 participants into two equal sets, DS (n = 25) and controls (n=25), and assessed. While a standard cephalometric analysis was conducted to measure the hard tissue parameters, tympanometry was used to assess the audiological parameters. The values of the linear and angular cephalometric parameters of the DS group were found to be lower than the controls. All the controls recorded type A tympanogram while the DS group recorded type A, B, and C tympanograms. A significant relationship was observed in the cephalometric readings – eustachian tube (ET) length, posterior upper facial height (PUFH) length, sella (s)-basion (ba)-palatal length (PL), and s-ba-ET angles – among the subjects who presented with type B or C tympanogram in comparison to those with type A. Tympanometry, a highly sensitive and relatively simple test to assess audiological parameters, has a significant relationship with a number of cephalometric indicators of growth and development. A deviation from the normal tympanometric readings can be used as an early indicator of the impending craniometric aberrations and handicap. This can be used as an effective tool for early intervention in cases of DS. Patients who have recorded abnormal tympanograms on multiple occasions over a period of six months can be subjected to a further cephalometric analysis.

Background: The importance of oral heath in the general wellbeing of our health is still largely undervalued. The current study was designed to comprehend the details of chief complaints and self-medication of the patients visiting a dental hospital in Mumbai. The level of awareness of the patients and its association with the practice of self-medication was also explored in the study. Methods: 300 randomly selected patients reporting to Nair Hospital Dental College were surveyed to conduct this clinical, observational, cross sectional, single center study. A specially designed and validated questionnaire consisting of 14 close ended questions was provided to the patients in the language of their convenience. Results: The most common chief complaint recorded among the dental patients was dental pain. This was followed by swelling, mobility and tooth replacement. The patients reported that most of them had experienced multiple dental problems before and did not report to the dental healthcare provider. 81% of the patients reported that they had practiced some form of self-medication for dental problems. 56.33% of them used analgesics, the pharmacy and over the counter drugs were the most common sources. A number of reasons were reported for the use of these drugs and the most common reason was to get quick or temporary relief but the patients did not report to the health care provider unless the symptoms got severe. Finally a significant relationship was recorded between the level of awareness of patients about the ill effects of self- medication and the prevalence of its practice among dental patients. Conclusions: More attention should be drawn towards the big impact of dental and oral health on the general well-being of the patients. The ethical use of medication should be encouraged not only by ensuring stringent rules governing the distribution of drugs but also by educating patients about the numerous risks involved with the practice of self-medication.