Problem formulation by medical students: an observation study

BACKGROUNDTest anxiety is prevalent among medical students and leads to impaired academic performance. Test-related attentional bias has been identified as an important maintaining factor in test-anxious individuals.AIMTo evaluate whether hypnosis and progressive muscle relaxation (PMR) could modify medical college students’ test anxiety and attentional bias.METHODSA total of 598 medical students were screened. The participants were divided into higher and lower test anxiety groups according to their scores on the test anxiety scale (TAS). Ninety medical college students with high TAS score were randomly assigned to a hypnosis or PMR group. Another 45 students with low TAS score were included, forming a baseline control group. The intervention was conducted weekly for 6 wk, and each session lasted approximately 30 min. The total intervention time and the number of intervention sessions for the hypnosis and PMR groups were equal. Data were collected at the pretest, posttest, and 2-mo follow-up.RESULTSHypnosis group participants had a significantly lower TAS score at posttest (t = -21.827, P < 0.001) and at follow-up (t = -14.824, P < 0.001), compared to that at pretest. PMR group participants also had a significantly lower TAS score at posttest (t = -10.777, P < 0.001) and at follow-up (t = -7.444, P < 0.001), compared to that at pretest. At the posttest level, the hypnosis group had a significantly lower TAS score than the PMR group (t = -3.664, P < 0.001). At the follow-up level, the hypnosis group also had a significantly lower TAS score than the PMR group (t = -2.943, P = 0.004). Clinically significant improvement was found in both the hypnosis and PMR groups (hypnosis = 64.0%; PMR = 62.22%). Hypnosis was more effective than PMR in reducing test anxiety among medical college students. Hypnosis could modify attentional bias toward threatening stimuli, but PMR could not.CONCLUSIONThese results suggest that attentional bias plays an important role in test anxiety treatment.

Diagnostic pitfalls and complex cases in MS.

Failure is an inevitable feature of innovation, and management research promulgates the importance of learning from it. Key to excelling at an innovation‐based strategy is understanding the processes that can turn failures into successes. However, post‐failure success remains elusive. Although failure signals that the innovation journey is off course, shifting trajectory is difficult, because it may require revising assumptions and reformulating the project’s problem representation. Using comparative case studies, this study set out to understand how problem representations are reformulated. Employing case method and comparing data versus theory iteratively, the important role of sensemaking and of leadership behaviors in driving post‐failure success became salient. Findings show that problem representations post‐failure require a process of problem formulation characterized by sensemaking and that innovative solutions are enabled by the reformulation of problem representations that spring from prospective sensemaking. Furthermore, this article identifies leadership change behavior as the linchpin driving a problem formulation process characterized by prospective sensemaking that catalyzes innovative solutions and explains why some projects thrive post‐failure and others do not. This article provides empirical support to the theoretical work of the literature on problem formulation, while extending the learning‐from‐failure literature by emphasizing and demonstrating the process driving post‐failure success. The major implication of our study is that different leadership behaviors may foster different types of sensemaking (retrospective or prospective), and that, in turn, the type of sensemaking matters for how a problem is reformulated. Ultimately, this article concludes that in the context of project failure, problem reformulation that springs from prospective sensemaking enables innovative solutions post‐failure.

Objective-Persons with severe mental illness have higher rates of chronic general medical illness compared with the general population.Similarly, compared with the general population, incarcerated persons have higher rates of chronic medical illness; however, there is little information about the synergy between severe mental illness and incarceration and the general medical problems of consumers.To address this gap in the literature this study addresses the following question: are consumers with a history of incarceration at greater risk of general medical problems compared with consumers without such a history?Methods-Administrative data were used to compare the medical problems of 3,690 persons with severe mental illness with a history of incarceration and 2,042 persons with severe mental illness with no such history.Results-Consumers with a history of incarceration were more likely than those with no such history to have infectious, blood, and skin diseases and a history of injury.Furthermore, when analyses controlled for gender, race, age, and substance use disorders, consumers with an incarceration history were 40% more likely to have any general medical problem and 30% more likely to have multiple medical problems.Conclusions-The findings presented here call for better communication among local public health and mental health providers and jails and better integration of primary care and behavioral health care among community mental health providers.Also, research on evidence-based interventions designed to divert persons with severe mental illness from the criminal justice system and facilitate community reentry for persons with severe mental illness who are released from jails and prisons should be accelerated.

Evaluate the impact of problem representation (PR) characteristics on Generative Artificial Intelligence (GAI) diagnostic accuracy. Internal medicine attendings and residents from two academic medical centers were given a clinical vignette and instructed to write a PR. Deductive content analysis described the characteristics comprising each PR. Individual PRs were input into ChatGPT-4 (OpenAI, September 2023) which was prompted to generate a ranked three-item differential. The ranked differential and the top-ranked diagnosis were scored on a 3-part scale, ranging from incorrect, partially correct, to correct. Logistic regression evaluated individual PR characteristic's impact on ChatGPT accuracy. For a three-item differential, accuracy was associated with including fewer comorbidities (OR 0.57, p=0.010), fewer past historical items (OR 0.60, p=0.019), and more physical examination items (OR 1.66, p=0.015). For ChatGPT's ability to rank the true diagnosis as the single-best diagnosis, utilizing temporal semantic qualifiers, more semantic qualifiers overall, and adhering to a typical 3-part PR format all correlated with diagnostic accuracy: OR 3.447, p=0.046; OR 1.300, p=0.005; OR 3.577, p=0.020, respectively. Several distinct PR factors improved ChatGPT diagnostic accuracy. These factors have previously been associated with expertise in creating PR. Future studies should explore how clinical input qualities affect GAI diagnostic accuracy prospectively.

Towards a goal-oriented methodology for clinical-guideline-based management recommendations for patients with multimorbidity: GoCom and its preliminary evaluation.

Diagnostic competence in students is a major medical education goal. Adding instructional guidelines to prompt deliberate reflection fosters medical students' diagnostic proficiency. This study investigates the effects of this teaching strategy on diagnostic accuracy in solving clinical cases of different complexity levels by novice and senior students. Eighty third-year and 62 sixth-year medical students participated in this three-phase experimental study. First, participants were randomly assigned to one of three experimental conditions (free reflection, cued reflection and worked example) to diagnose 12 clinical text-based cases, following different levels of deliberate reflection. In an immediate test and a delayed test, the participants diagnosed varied sets of 12 cases, six involving the same diseases (four routine and two rare). The main outcomes were the diagnostic accuracy scores achieved for the cases assessed by repeated measures of analysis of variance for each category. There was a significant primary effect of experimental condition (P<.001), year of training (P<.001) and study phase (P<.001) on the diagnostic accuracy achieved. The use of deliberate reflection in addition to instructional guidelines resulted in improved results in the immediate test for all cases evaluated (P<.001), regardless of participants' seniority. In the delayed test, this benefit was maintained for simple cases (P<.001). For complex cases, the benefit was maintained only for senior students (P<.001). The cued reflection and worked example groups did not differ in performance (P>.05), but both groups surpassed the free reflection group (P<.001), regardless of the students' learning stage and case complexity.

BackgroundThe exponential growth of medical knowledge presents a paradox for modern medical education. While access to information is immediate, applying it in a clinically meaningful way remains a challenge. Large language models (LLMs), such as ChatGPT, are widely used for information retrieval, yet their role in dynamic, high-pressure clinical learning remains poorly understood.ObjectiveThis study aims to evaluate whether unstructured access to an LLM improves decision-making, teamwork, and confidence in trauma education for medical students.MethodsThis randomized controlled pilot study involved 41 final-year medical students participating in a trauma simulation session. Students self-selected into teams of 4 to 6 and were randomized to either an LLM-assisted group (ChatGPT-4o mini) or a control group without LLM access. All teams completed 18 video-based trauma scenarios requiring time-sensitive clinical decisions. Prompting was unrestricted. Confidence and trauma exposure were assessed using pre- or postquestionnaires. Facilitators rated teamwork (1-5), decision accuracy, and response times. Knowledge retention was measured 4 weeks later via an online quiz.ResultsConfidence in trauma management improved in both groups (P<.001), with larger gains in the non-LLM group (P=.02). LLM support did not enhance the decision accuracy or speed and was associated with longer response times in some complex cases. Teams without LLMs demonstrated more active discussion and scored higher in teamwork ratings (median 5.0 [IQR 5.0-5.0] vs median 3.5 [IQR 3.0-4.5]; P=.08). Students primarily used the LLM for fact-checking but reported vague or overly general responses. Knowledge retention was high across both groups and did not differ significantly (P=.33).ConclusionsWhile students appreciated the inclusion of artificial intelligence (AI), unstructured LLM use did not improve performance and may have disrupted the group reasoning. The use of non-English prompting likely contributed to lower AI performance, underscoring the importance of language alignment in LLM applications. This pilot study highlights the need for structured AI integration and targeted instruction in AI literacy. Simulation-based trauma education proved effective and well received, but optimizing the educational value of LLMs will require thoughtful curricular design. Further studies with more students are needed to define best practices for LLM use in clinical education.

The article aims to review current trends in the organization of self-education activities for medical college students, models of blended learning, and its main components that ensure the self-education of future nursing specialists. The research methods are terminological and comparative analysis, generalization, specification, and modelling. Based on the analysis of primary sources, it has been established that self-education activities are a type of educational and cognitive activity motivated by an internal drive for self-improvement and facilitated by the use of specific methods for seeking and assimilating social experience. It has been emphasized that through self-education, medical professionals are better able to handle complex clinical cases and provide high-quality medical care to their patients, including via the use of digital technologies. Blended learning is defined as a combination of traditional face-to-face (contact) learning and computer-mediated or online learning. It has been pointed out that the implementation of blended learning requires the development of appropriate pedagogical design and innovations that ensure the application of adequate teaching and assessment methods with carefully planned content and structured academic courses. The blended learning system is presented as an integration of face-to-face classroom and clinical sessions, independent student work, and online learning. It has been noted that in blended models, the supply of educational material and interaction with the teacher can occur in both physical and virtual environments, during both classroom sessions and asynchronously during extracurricular activities. The specifics of self-education activities for medical college students using some models such as rotational, laboratory, independent blended learning, flexible learning, enriched virtual learning, flipped classroom, and personalized learning have been described. In each model, the core element is identified as the electronic educational resource. In the blended learning system such a resource is viewed not as an auxiliary tool of the traditional educational process but as the one that successfully combines traditional and distance learning forms, incorporating both non-digital and digital technologies. It has been determined that blended learning is based on the flexible combination of in-class learning (contact hours) with online sessions (asynchronously or synchronously). Among the tools for blended learning used in organizing self-education activities for medical college students, the following have been highlighted and described: online lectures, practical sessions in webinar mode, project and group work online, learning management systems such as LMS Moodle, laboratory work in a blended format, virtual and augmented reality technologies, a combination of formal and nonformal education, self-testing complexes, webinar consultations, artificial intelligence technologies, and specialized medical applications. It has been concluded regarding the feasibility of implementing blended learning in the training of medical college students and the organization of their self-education activities. This will enhance the quality of education, and the efficiency of resource utilization of educational institutions, increase the level of professional expertise and mobility of teachers, and ultimately provide additional competitive advantages to the educational institution in the educational services market.

• On the basis of class B evidence and consensus, acute appendicitis in children can often be diagnosed clinically with only selective use of imaging. (13)(14)(15)(16) • On the basis of class B evidence and consensus, ultrasonography is the test of choice when acute appendicitis is suspected but is unclear based on history, physical examination, and laboratory results. (17)(18)(19) • On the basis of class B evidence and consensus, the use of computed tomography scan should be limited to cases of suspected complex appendicitis with abscess or when there is clinical suspicion for acute appendicitis but ultrasonography results are not helpful. (16) • On the basis of class C evidence and consensus, children with possible appendicitis ideally should be treated in medical centers that have skilled sonographic personnel. (21) • On the basis of class B evidence and consensus, simple appendicitis should be treated by appendectomy during normal operating hours. Preoperative treatment with intravenous antibiotics and fluids during the overnight hours halts disease progression and allows for the safest surgery with the benefit of a full and rested staff. (24)(25)(26) • On the basis of class B evidence and consensus, complex appendicitis with a well-defined abscess can be treated nonoperatively initially, with the option of an interval appendectomy after recovery from the acute infection. (29)(30) (31)(32)(33)

BackgroundHistory-taking is crucial in medical training. However, current methods often lack consistent feedback and standardized evaluation and have limited access to standardized patient (SP) resources. Artificial intelligence (AI)–powered simulated patients offer a promising solution; however, challenges such as human-AI consistency, evaluation stability, and transparency remain underexplored in multicase clinical scenarios.ObjectiveThis study aimed to develop and validate the AI-Powered Medical History-Taking Training and Evaluation System (AMTES), based on DeepSeek-V2.5 (DeepSeek), to assess its stability, human-AI consistency, and transparency in clinical scenarios with varying symptoms and difficulty levels.MethodsWe developed AMTES, a system using multiple strategies to ensure dialog quality and automated assessment. A prospective study with 31 medical students evaluated AMTES’s performance across 3 cases of varying complexity: a simple case (cough), a moderate case (frequent urination), and a complex case (abdominal pain). To validate our design, we conducted systematic baseline comparisons to measure the incremental improvements from each level of our design approach and tested the framework’s generalizability by implementing it with an alternative large language model (LLM) Qwen-Max (Qwen AI; version 20250409), under a zero-modification condition.ResultsA total of 31 students practiced with our AMTES. During the training, students generated 8606 questions across 93 history-taking sessions. AMTES achieved high dialog accuracy: 98.6% (SD 1.5%) for cough, 99.0% (SD 1.1%) for frequent urination, and 97.9% (SD 2.2%) for abdominal pain, with contextual appropriateness exceeding 99%. The system’s automated assessments demonstrated exceptional stability and high human-AI consistency, supported by transparent, evidence-based rationales. Specifically, the coefficients of variation (CV) were low across total scores (0.87%‐1.12%) and item-level scoring (0.55%‐0.73%). Total score consistency was robust, with the intraclass correlation coefficients (ICCs) exceeding 0.923 across all scenarios, showing strong agreement. The item-level consistency was remarkably high, consistently above 95%, even for complex cases like abdominal pain (95.75% consistency). In systematic baseline comparisons, the fully-processed system improved ICCs from 0.414/0.500 to 0.923/0.972 (moderate and complex cases), with all CVs ≤1.2% across the 3 cases. A zero-modification implementation of our evaluation framework with an alternative LLM (Qwen-Max) achieved near-identical performance, with the item-level consistency rates over 94.5% and ICCs exceeding 0.89. Overall, 87% of students found AMTES helpful, and 83% expressed a desire to use it again in the future.ConclusionsOur data showed that AMTES demonstrates significant educational value through its LLM-based virtual SPs, which successfully provided authentic clinical dialogs with high response accuracy and delivered consistent, transparent educational feedback. Combined with strong user approval, these findings highlight AMTES’s potential as a valuable, adaptable, and generalizable tool for medical history-taking training across various educational contexts.

This article offers a comparative study between two research fields in the sociology of work: occupational health and safety services, and housing renovation associations combating substandard housing. Although these contexts appear distant, they share a common mechanism: professionals’ orientation towards the management of “complex cases.” By complex cases, we mean situations that escape established professional routines, require the mobilization of diverse knowledge, and strongly engage professional responsibility. The “Dr. House effect” illustrates how occupational health physicians find meaning and legitimacy in resolving difficult medical problems, while the “Maçons du cœur effect” highlights how associations address housing renovation for highly precarious households. These popular culture references emphasize how complexity becomes both a source of meaning and a lever for professional legitimation. However, while occupational physicians, an established and legally protected profession, reinforce their dominant position through this specialization, associations, situated in a subordinate position within their field, struggle to stabilize their recognition. The analysis thus sheds light on the contrasting professional dynamics between established professions and emerging associative actors.

Clinical decision making requires knowledge, experience and analytical/non-analytical types of decision processes. As clinicians progress from novice to expert, research indicates decision-making becomes less reliant on foundational biomedical knowledge and more on previous experience. In this study, we investigated how knowledge and experience were reflected in terms of differences in neural areas of activation. Novice and expert clinicians diagnosed simple or complex (easy, hard) cases while functional magnetic resonance imaging (fMRI) data were collected. Our results highlight key differences in the neural areas activated in novices and experts during the clinical decision-making process. fMRI data were collected from ten second year medical students (novices) and ten practicing gastroenterologists (experts) while they diagnosed sixteen (eight easy and eight hard) clinical cases via multiple-choice questions. Behavioral data were collected for diagnostic accuracy (correct/incorrect diagnosis) and time taken to assign a clinical diagnosis. Two analyses were performed with the fMRI data. First, data from easy and hard cases were compared within respective groups (easy>hard, hard>easy). Second, neural differences between novices and experts (novice>expert, expert>novice) were assessed. Experts correctly diagnosed more cases than novices and made their diagnoses faster than novices on both easy and hard cases (all p's<0.05). Time taken to diagnose hard cases took significantly longer for both novices and experts. While similar neural areas were activated in both novices and experts during the decision making process, we identified significant hemispheric activation differences between novice and expert clinicians when diagnosing hard clinical cases. Specifically, novice clinicians had greater activations in the left anterior temporal cortex and left ventral lateral prefrontal cortex whereas expert clinicians had greater activations in the right dorsal lateral, right ventral lateral, and right parietal cortex. Hemispheric differences in activation were not observed between novices and experts while diagnosing easy clinical cases. While clinical decision-making engaged the prefrontal cortex (PFC) in both novices and experts, interestingly we observed expertise related differences in the regions and hemispheres of PFC activation between these groups for hard clinical cases. Specifically, in novices we observed activations in left hemisphere neural regions associated with factual rule-based knowledge, whereas in experts we observed right hemisphere activation in neural regions associated with experiential knowledge. Importantly, at the neural level, our data highlight differences in so called type 2 clinical decision-making processes related to prior knowledge and experience.

Oral assessments are essential components of board certification in numerous fields, as they provide insight into problem-solving capacity and clinical reasoning. The development of clinical reasoning often begins in undergraduate medical education and remains a challenge to assess. We developed a pilot oral assessment to evaluate medical student oral presentations and systematically assess clinical reasoning. This was incorporated into a previously existing cumulative assessment at the conclusion of the third year of medical school, with the intent to demonstrate feasibility and future reliability of this exam format. This pilot oral assessment was developed using content taught during third year clerkships. A modified Assessment of Reasoning Tool (ART) was used as the evaluation metric. It was conducted virtually to include faculty members from multiple disciplines and accommodate schedules and space limitations. A total of 152 third year medical students completed the exam, with a total of 15 faculty examiners. 89% of medical students scored as complete in hypothesis directed history, 93% in problem representation, 86% in prioritized differential diagnoses, and 67% in effectively directing management. Most examiners felt an oral assessment is effective to determine a medical student's clinical reasoning ability. Virtual oral assessments of clinical reasoning can be incorporated in undergraduate medical education to identify students struggling with components of clinical reasoning, while also allowing maximum flexibility for the clinician educator workforce as examiners. Longitudinal use of these exams would be valuable to track the development of clinical reasoning across the medical school curriculum.

BackgroundLarge language models (LLMs) have demonstrated potential in assisting clinical decision-making. However, studies evaluating LLMs’ diagnostic performance on complex critical illness cases are lacking. We aimed to assess the diagnostic accuracy and response quality of an artificial intelligence (AI) model, and evaluate its potential benefits in assisting critical care residents with differential diagnosis of complex cases.MethodsThis prospective comparative study collected challenging critical illness cases from the literature. Critical care residents from tertiary teaching hospitals were recruited and randomly assigned to non-AI-assisted physician and AI-assisted physician groups. We selected a reasoning model, DeepSeek-R1, for our study. We evaluated the model’s response quality using Likert scales, and we compared the diagnostic accuracy and efficiency between groups.ResultsA total of 48 cases were included. Thirty-two critical care residents were recruited, with 16 residents assigned to each group. Each resident handled an average of 3 cases. DeepSeek-R1’s responses received median Likert grades of 4.0 (IQR 4.0–5.0; 95% CI 4.0–4.5) for completeness, 5.0 (IQR 4.0–5.0; 95% CI 4.5–5.0) for clarity, and 5.0 (IQR 4.0–5.0; 95% CI 4.0–5.0) for usefulness. The AI model’s top diagnosis accuracy was 60% (29/48; 95% CI 0.456–0.729), with a median differential diagnosis quality score of 5.0 (IQR 4.0–5.0; 95% CI 4.5–5.0). Top diagnosis accuracy was 27% (13/48; 95% CI 0.146–0.396) in the non-AI-assisted physician group versus 58% (28/48; 95% CI 0.438–0.729) in the AI-assisted physician group. Median differential quality scores were 3.0 (IQR 0–5.0; 95% CI 2.0–4.0) without and 5.0 (IQR 3.0–5.0; 95% CI 3.0–5.0) with AI assistance. The AI model showed higher diagnostic accuracy than residents, and AI assistance significantly improved residents’ accuracy. The residents’ diagnostic time significantly decreased with AI assistance (median, 972 s; IQR 570–1320; 95% CI 675–1200) versus without (median, 1920 s; IQR 1320–2640; 95% CI 1710–2370).ConclusionsFor diagnostically difficult critical illness cases, DeepSeek-R1 generates high-quality information, achieves reasonable diagnostic accuracy, and significantly improves residents’ diagnostic accuracy and efficiency. Reasoning models are suggested to be promising diagnostic adjuncts in intensive care units.