United States

Background
Type 2 diabetes mellitus (T2DM) is a global epidemic affecting approximately 9.3% of the world population and has an all-cause mortality rate of 8.5%. An increasing number of patients with chronic conditions access online resources for educational purposes, with recent statistics suggesting that up to 80% of adults in the United States have obtained medical information through online search engines. This study aims to assess the accuracy of the information offered by artificial intelligence (AI) chatbots to T2DM questions using ChatGPT.

Methods:
An initial set of questions about T2DM and its complications were obtained from an online platform such as American Diabetes Association. Subsequently, these questions underwent a thorough evaluation by physicians well-versed in diabetes management before being presented to ChatGPT. Each question was introduced to ChatGPT three times, and the responses were recorded. Two study staff reviewed the three sets of responses and summarized the majority response that was repeated at least twice. Data were reviewed by two board-certified internal medicine physicians with a minimum of 11 years of experience. Responses were categorized into appropriate; provide accurate information that meets the standard of care, inappropriate; contain erroneous information but meets the minimal standard of care, and unreliable; the response provides inconsistent content that does not meet the standard of care.


Results:
Of the original 110 questions, 70 were included in the final dataset. Out of the 70 summarized responses generated by ChatGPT, 89.3% were categorized as appropriate, indicating that the AI model provided accurate information meeting the standard of care for Type 2 diabetes management and its associated complications. On the other hand, 10% of the responses were deemed inappropriate, containing errors but still meeting the minimal standard of care in addressing the questions. The remaining 0.7% of ChatGPT&#39;s responses were classified as unreliable, displaying inconsistent content that did not meet the standard of care in terms of accuracy and reliability


Conclusion
ChatGPT offers a promising source of information for educating patients with T2DM. The study demonstrates high accuracy in providing appropriate responses that align with the standard of care, positioning it as a useful tool in enhancing patient education and diabetes management.


AMA Research Challenge 2023

Artificial Intelligence Shifting the Paradigm of Diabetes Care

Background
Type 2 diabetes mellitus (T2DM) is a global epidemic affecting approximately 9.3% of the world population and has an all-cause mortality rate of 8.5%. An increasing number of patients with chronic conditions access online resources for educational purposes, with recent statistics suggesting that up to 80% of adults in the United States have obtained medical information through online search engines. This study aims to assess the accuracy of the information offered by artificial intelligence (AI) chatbots to T2DM questions using ChatGPT.

Methods:
An initial set of questions about T2DM and its complications were obtained from an online platform such as American Diabetes Association. Subsequently, these questions underwent a thorough evaluation by physicians well-versed in diabetes management before being presented to ChatGPT. Each question was introduced to ChatGPT three times, and the responses were recorded. Two study staff reviewed the three sets of responses and summarized the majority response that was repeated at least twice. Data were reviewed by two board-certified internal medicine physicians with a minimum of 11 years of experience. Responses were categorized into appropriate; provide accurate information that meets the standard of care, inappropriate; contain erroneous information but meets the minimal standard of care, and unreliable; the response provides inconsistent content that does not meet the standard of care.


Results:
Of the original 110 questions, 70 were included in the final dataset. Out of the 70 summarized responses generated by ChatGPT, 89.3% were categorized as appropriate, indicating that the AI model provided accurate information meeting the standard of care for Type 2 diabetes management and its associated complications. On the other hand, 10% of the responses were deemed inappropriate, containing errors but still meeting the minimal standard of care in addressing the questions. The remaining 0.7% of ChatGPT's responses were classified as unreliable, displaying inconsistent content that did not meet the standard of care in terms of accuracy and reliability


Conclusion
ChatGPT offers a promising source of information for educating patients with T2DM. The study demonstrates high accuracy in providing appropriate responses that align with the standard of care, positioning it as a useful tool in enhancing patient education and diabetes management.


poster

Welcome to the **largest national, multi-specialty research** event featuring research from nearly 700 medical students, residents, fellows, and international medical graduates.

Use the buttons below or navigation menu to access the event.
 


[![](https://assets.underline.io/uploads/markdown_image/1/image/3e33a8c754cc49610b30950617d813e7.png)](https://underline.io/events/434/sessions?searchGroup=all)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/551404d429c4e274f9a54cbdd7081918.png)](https://underline.io/events/434/posters?searchGroup=all)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/677d92f566b880cebd36905fc738fc10.png)](https://underline.io/events/434/lounge)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/c7392833c7261c3874c5b43fa2f4df19.png)](https://underline.io/events/434/help-desk)
 

#### Event Details

The 2023 American Medical Association Research Challenge includes a poster symposium hall and a semifinals hall with the top scored research. **Log in at any time throughout the event to:**

* **View posters and presentations** on a variety of topics and specialties
* **Score posters in the semifinals** to help decide the top five finalists (AMA member exclusive opportunity)
* **Connect with participants** and provide feedback to help advance their research
* **Attend education sessions** with physician experts on key research topics

At the conclusion of this event, the AMA will announce the top five finalists who will advance to the AMA Research Challenge finals, present their research to an elite panel of judges, and compete to win a $10,000 grand prize sponsored by Laurel Road. 

### Grand Prize Sponsor

AMA would like to thank Laurel Road for sponsoring a $10,000 grand prize for the winner of the AMA Research Challenge. Laurel Road is not involved in the judging or scoring process to select the winner. The winner will be announced on February 4, 2024 during the AMA Research Challenge finals. Payment of the grand prize will be directly from Laurel Road. See the AMA Research Challenge Official Rules for terms and conditions.

![](https://assets.underline.io/uploads/markdown_image/1/image/611e331d4a0eba0920f240cfc04b4d77.png)

**There is no registration fee for this event! **

Register to explore research in a variety of topics. AMA members get exclusive access to score posters in the semifinals.

**Please use this link [here](https://www.ama-assn.org/about/events/ama-research-challenge) to register for free.**

Registration Is Required

Join the AMA Research Challenge poster symposium and semifinals to explore top research posters, score posters in the semifinals (an AMA member exclusive opportunity) and attend education sessions with experienced physicians to learn tips to advance your research and career. 

Background
Hypertension is the number one preventable risk factor for all-cause mortality, affecting up to 45% of the adult population worldwide. In the era of the growing influence of artificial intelligence (AI), reliance on online platforms for medical information is more than ever before, resulting in concerns surrounding the accuracy of the available information. This study aims to assess the accuracy of text-generating AI using ChatGPT in answering patients’ frequently asked questions about hypertension and its complications.
methods
To assess the accuracy of the information offered by ChatGPT on hypertension and secondary complications, this nonhuman-subject qualitative study was designed. Frequently asked questions were gathered by study staff from professional platforms, including the American Heart Association. These questions were then reviewed by two physicians experienced in treating hypertension, and a final set of questions was established. Each question was put through the ChatGPT three times, and the majority- response to each question was evaluated against the national guidelines by study staff who graduate physicians in internal medicine. A board-certified internal medicine physician with over 8 years of experience then reviewed all questions and categorized responses in one of the following 3 classes: Responses were classified into three categories by accuracy: appropriate (in line with clinical recommendations), inappropriate (contain errors), or unreliable (inconsistent). Descriptive statistical analysis was performed to compare ChatGPT responses against the standard and assessed them for accuracy and reproducibility.
Results
Of the total of 130 questions initially proposed, a total of 100 questions were deemed appropriate and were selected for the purpose of this study. When assessed against acceptable standard responses, ChatGPT responses were found to be appropriate in 91.5% of cases, inappropriate in 0.5%, and unreliable in 8%. Notably, out of 300 responses, 4% were irreproducible, meaning that they could not be consistently reproduced across multiple ChatGPT runs.
Conclusion
ChatGPT demonstrated promising accuracy in answering frequently asked questions related to hypertension. These findings highlight the potential for AI chatbots, such as ChatGPT, in offering valuable

information to patients regarding hypertension. Further research and refinement are needed to ensure the reliability and broader applicability of ChatGPT in the medical field.

Assessing the Accuracy and Reproducibility of Artificial Intelligence Chatbot in Answering Frequently Asked Questions Regarding Hypertension and its Associated Complications

Background
Neurosurgery residency is a fast-paced, time-consuming entity, which results in residents commonly approaching, if not violating the 80-hour workweek limit. This involves a complex structure with multiple hospitals, services, and clinic days, leading to challenges in creating fair call schedules. Historically, the manually-prepared scheduling system by the chief resident has been prone to biases, mistakes, and perceived unfairness. To address these issues, an automated and optimized neurosurgery scheduling system (AONSS) has been developed to reduce biases, accommodate resident requests, and optimize call variation, ultimately enhancing the educational experience by promoting diverse junior-senior-attending relationships. Of note, the national neurosurgery residency pool is reported to be ~18% female. 

Methods
AONSS was built using a combination of Perl and Visual Basic. The inaugural use of the AONSS occurred on July 1, 2021, in an optional manner, to the chiefs during the 2021-2022 academic year. An updated version of the program was implemented July 1, 2022 for the 2022-2023 academic year. Residents were surveyed using Google Forms with anonymous responses prior to its implementation and again as a follow up in July 2023. Objective primary outcome measure: senior and junior call variation (variance); Objective secondary outcome measures: variation in senior and junior Saturday call, senior and junior weekend call, senior-junior pairings, and resident-attending pairings. Subjective primary outcome measure: perceived satisfaction score based on 5-point Likert scale; Subjective secondary outcome measures: perceived fairness score (5-point Likert scale), perceived gender-bias (dichotomous) and perceived racial-bias (dichotomous).

Results
26 residents (28-39% female/year) were included in the study. AONSS was used for 6/13 blocks during the 2021-2022 academic year and 13/13 blocks for the 2022-2023 academic year. Overall call variation reduced by 70%. All other objective secondary measures have improved with AONSS. Monthly duty hours were reduced and less varied. Satisfaction scores improved from 21% reporting being somewhat satisfied or very satisfied to 90%. Fairness scores improved from 43% reporting being somewhat fair or very fair to 95%. Perception of gender bias decreased from 29% to 0%. No resident felt there was racial bias in either system.

Conclusion

Our newly developed automated scheduling system effectively reduces variation among calls in the complex UF neurosurgery residency, which in return, was found to increase residents’ satisfaction with their schedule, improve their perception of fairness with the schedule, and has completely removed the perception of sexual bias in a program that has a large percentage of females. Additionally, it was found to be associated with a decrease in duty hours. 


Automated and Optimized Neurosurgery Scheduling System Improves Resident Satisfaction

Background: Thrombolytic therapy is the standard of care in eligible patients presenting with acute ischemic stroke (AIS). Physicians are challenged to: 1) identify an AIS; 2) explain thrombolytic therapy; and 3) consent the patient and/or surrogate decision-maker efficiently and effectively. There is no consensus in clinician practice towards obtaining and documenting consent for thrombolytics in AIS. The objective of this study was to determine patient demographics and clinical characteristics that impact whether consent was documented for administration of thrombolytic therapy for AIS.

Methods: A multicenter mixed-methods cohort study included 442 patients who received thrombolytics for AIS. Demographics, including primary language, were collected from the electronic medical record (EMR). Code status, co-morbid dementia, stroke location, and initial NIHSS were also collected. Chi Square analysis was conducted to determine differences between two cohorts, patients who received thrombolysis and had documented consent and those who did not have documented consent in their EMR. Bivariate and multivariate regression were performed to determine associations with having documented consent. Thematic analysis was conducted on documented conversations to identify themes regarding consent for thrombolytics.

Results: The minority of patients (28.5%) had documented consent for thrombolysis in their EMR. In multivariate analysis, patients who were older (p<0.001), non-white (p=0.004), primary language other than English (p=0.03), co-morbid dementia (p<0.001), or had a DNR/partial code status (p=0.02) were more likely to have documented consent. Thematic analysis identified themes for documenting consent including risks and benefits, outcomes, and paternalism.

Conclusions: Documented consent was present in the minority of EMRs for patients treated with thrombolytics for AIS. Documented consent was more prevalent in populations with dementia, older age, primary language other than English, non-white, and code status. Qualitative analysis found effective consent was not documented in most cases. Given variation in consent documentation, it may be beneficial to develop standardized evidence-based language to guide these conversations.

Documentation of Consent for Thrombolytic Therapy in Acute Ischemic Stroke

Stress plays a central role in functioning for all life forms. As humans, we experience stress in a multitude of ways through various types of stimuli. Due to the constancy of stressors in our lives, the nervous system has learned to allosterically adapt to the stimuli. However, when the body cannot adapt to chronic stress it can have morphological and degenerative impacts on neuroanatomy and cognitive function that may or may not be reversible. In this literature review, our aim was to identify the specific neuroanatomical structures that were impacted most by the long-term effects of chronic stress. We also describe the impact the subsequent relationship between chronic stress and the morphological changes had on cognitive function, especially on medical students. We completed our analysis through a compilation of clinical trials and other systematic reviews. Our sources were assessed on credibility and referenced from Google Scholar, PubMed, Science Direct, and NCBI. The effects of various types of chronic stress had the most impact on neurogenesis and neuronal development, brain plasticity, and spatial learning and memory with association to the hippocampus. These structures had both reversible and non-reversible morphological alterations, which also had a direct impact on the brain’s cognitive abilities. While the results and impacts of chronic stress are still being conducted, future research may be able to further highlight why stressful stimuli can particularly impact these structures and the tangential impacts that it may have on related or adjacent structures. 
Introduction
With the hustle and bustle of daily life and the constant pressures of the evolving world, chronic stress has become a commonplace occurrence. Stress is categorized as any intrinsic or extrinsic stimulus that evokes a biological response, with the compensatory responses to these stimuli labeled as stress responses (1). Stressors can have an influence and impact on psychosocial health, physiological health, and even how the body maintains homeostasis (2). While the central nervous system is primed to produce integrated coping responses, it is the autonomic nervous system that engages in the fight, flight, or coping mechanisms that help the body adjust and adapt to the stressful stimuli that it may encounter (3). Some stress responses may be instantaneous, but if the stimuli persist over a long period of time, it is generally referred to as chronic stress (3). 
How an individual reacts to the strain that long-exposed stimuli pose can potentially affect both mental and physical health, with chronic stress having a more permanent effect on the body than short term stressors (4). Additionally, the time course of stress responses can be measured by the neuroendocrine and behavioral responses from the body, which can identify how destabilizing the stimulus makes the stress manageable or harmful (1). How the stressors are processed, in turn, can have a remarkable toll on the development and functioning of the brain and nervous systems. 
The brain plays a key integratory role in processing stimuli in the adaptation to stress. Stress and stimuli are information that stimulate behavioral and physiological responsiveness between the brain and other systemic functions (4). Beyond the acute responses to stress that life may pose, allostatic load is considered to be the integral feature of chronic stress that leads to degenerative effects (5). Allostatic load includes innate and adaptive processes that help to maintain homeostasis through mediator chemical messengers like adrenaline and cortisol (6). With allostatic load comes the heightened wear and tear on the central and peripheral nervous systems and even down to the cellular level (7, 8). This systematic review aims to focus on identifying key neurological structural changes that may be induced by the effects of chronic stress and put into context how these changes may, in turn, lead to changes in behavioral or mental health as well.


The Effects of Chronic Stress on Neuroanatomy and Cognitive Function

Background:
Pregnant people face higher morbidity and mortality from influenza (flu) and COVID-19 infection than non-pregnant counterparts. Despite high vaccine efficacy, only 50% of eligible pregnant people received flu vaccine before or during pregnancy and 22% completed COVID vaccination series (including boosters) as of January 2022 and December 2022, respectively.

Objective:
Qualitatively assess how providers and nurses recommend flu and COVID vaccines during prenatal visits in obstetric practices that participated in a flu vaccine communication randomized control trial (RCT) in Rochester, NY and Los Angeles, CA.

Methods:
34 obstetric practices (21 in NY, 13 in CA) participated in an RCT to increase flu vaccination rates. 16 intervention sites received a total of 60 minutes of online training on flu morbidity, vaccine effectiveness, motivational interviewing, discussing protection of the infant, and using presumptive recommendations (“you are due for a flu vaccine”). At the end of the study, provider and nurse participants received a REDCap survey with open ended questions about how they recommend flu and COVID vaccinations during prenatal visits. Data collection occurred from July - October 2022. Investigators applied content analysis to code qualitative data into themes. Results were stratified according to study arm.

Results:
The survey response rate was 46% (124/271); and 44% (104/239) of providers and 59% (19/32) of nurses participated. For flu vaccination, increased risk of flu during pregnancy was the most common theme noted in both study arms. Intervention sites more frequently described using presumptive recommendations (14% intervention, 3% control) and normalized vaccinating (“we recommend flu vaccine for all prenatal patients”, 30% intervention, 23% control) to encourage flu vaccination. Presumptive recommendation for COVID vaccination was infrequent and occurred exclusively in intervention sites (6%). Respondents from intervention sites more often mentioned protection of the baby when discussing COVID vaccination (46% intervention, 38% control).

Conclusion:
In a qualitative analysis of flu and COVID vaccine discussions for practices that participated in a flu vaccine communication RCT, participants in intervention sites more often used presumptive recommendations and normalized vaccinating, in addition to using skills taught for flu vaccination while discussing COVID vaccination. Next steps include formally testing similar interventions to increase COVID vaccination rates for pregnant people.


Provider and nurse recommendations for Influenza and COVID-19 vaccination in obstetric practices: a qualitative analysis

Background
Due to nationwide physician shortages, many medical schools across the United States have established regional medical campuses (RMCs) to increase class size. The University of Missouri-Columbia is one such institution, creating the Springfield Clinical Campus (SCC) for third- and fourth-year students, with the first class graduating in 2018. SCC students complete their core clerkships in two community hospitals, while their central campus peers complete clerkships in a university setting. At this time, no literature exists investigating whether RMC graduates are more likely than their central campus counterparts to train in community, university, or community-based university affiliated residency programs. In this preliminary study, we compare the types of obstetrics and gynecology residency (OBGYN) programs matched by RMC and central campus graduates at the University of Missouri-Columbia School of Medicine. 

Methods
We utilized publicly available OBGYN match data from 2018-2023 for the University of Missouri-Columbia. Each residency program matched was categorized in the Fellowship and Residency Electronic Interactive Database Access (FREIDA) as university-based, community-based, or community-based university affiliated as self-reported by the programs. The Freeman-Halton extension of the Fisher exact probability test was used to calculate the two-tailed p-value comparing the proportions of students who matched at community, university, and community-based university affiliated programs.

Results
In total, 34 students who matched OBGYN from the University of Missouri-Columbia elected to make their data public. Of these, 8 students attended the SCC and 26 students attended the central campus. At the SCC, 50% of students matched at university programs, 50% of students matched at community-based university affiliated programs, and 0 students matched at community programs. At the central campus, 50% of students matched at university programs, 31% of students matched at community-based university affiliated programs, and 19% of students matched at community programs. There was no statistical difference between the campuses for the proportions of students who matched at each type of OBGYN program (p=0.4067).

Conclusion
No statistical difference between the central campus and SCC was found for the proportions of students matching at university, community, and community-based university affiliated OBGYN programs. Subsequent research including a longer timespan may be useful for elucidating trends among students training at regional versus central campuses.

A Comparison of Regional Medical Campus and Central Campus Matches into 
Obstetrics and Gynecology Programs by Type

In recent years, there has been a decline in the rates of trabeculectomy and a rise in rates of aqueous shunts and microinvasive glaucoma surgeries. , There is concern amongst glaucoma surgical educators that low trabeculectomy volumes during training are exacerbating this trend. 
The purpose of this cross-sectional survey study was to survey recently graduated glaucoma specialists and explore how surgical volume during training, success and complication rates, practice setting, and other patient population factors might influence practice patterns of recently graduated glaucoma specialists. 
A survey was distributed to the American Glaucoma Society’s “Young Glaucoma Specialists” listserv, and 66 responses were received from members completing training within the past 5 years (2018-2022). 
Sixty four respondents (93.9%) reported feeling somewhat or entirely comfortable performing trabeculectomy as a result of their training. However, 42 (63.6%) said that they do not often perform trabeculectomy, whereas only 13 (19.7%) reported that they do. 
Respondents were offered a list of factors that might influence their choice to perform trabeculectomy and selected the degree to which they agreed with each. Those who reported performing trabeculectomy often (N = 13) were influenced by the surgery’s high success rates (92.3%), its preoperative (76.9%) and intraoperative (84.6%) processes, and low postoperative complications (61.5%). Those who reported not performing trabeculectomy often (N = 42) were most discouraged by the postoperative process of trabeculectomy, as well as the socioeconomic (66.7%) and sociocultural (52.4%) characteristics of their patients. Factors such as volume performed during training (47.6%), cost-effectiveness (19.0%), and reimbursement rates (19.0%) were not considerable contributors. 

While the majority of young glaucoma specialists in our study report being comfortable performing trabeculectomy as a result of their training, the majority stated that they do not often perform trabeculectomy in practice. Notably, low trabeculectomy surgical volume during training was not cited as an notable barrier, whereas patient factors and preference for alternative procedures, depending on the clinical scenario, appeared to drive this trend.

Young Glaucoma Specialist Practice Patterns: Why do you do what you do?

Background: Clinical exposure to ophthalmology during medical school is minimal, and required ophthalmology clinical rotations have declined from 60% to 16% of Association of University Professors of Ophthalmology (AUPO)-affiliated medical schools from 2000-2018. A study evaluating medical student perception of an ophthalmology career found that “insufficient interest” and “not enough exposure” are the top reasons for not pursuing ophthalmology. Thus, there is a need to facilitate interest and exposure through extracurricular activities, such as community service. This study investigates medical students’ motivation to volunteer at a free eye clinic, prior ophthalmology exposure, and specialty interest after volunteering at Storm Eye Institute’s Fight for Sight clinic, a student and resident-run free eye clinic. 
Methods: This cross-sectional study was conducted by an online RedCap survey sent to medical student volunteers after volunteering at Storm Eye Institute’s Fight for Sight clinics between February-April 2023. Through multiple choice, open-ended questions, and Likert scales, the survey asked about demographics, prior exposure to ophthalmology, motivation to volunteer at the clinic, experience at the clinic, and career exploration. 
Results: 12 out of 18 medical student volunteers responded to the survey, representing a 66.7% response rate. Demographics revealed that most participants identified as male (75% male, 25% female) and white (67% white, 25% Asian, 8% prefer not to answer). Average exposure to ophthalmology was 4.75/10 with 0 representing no exposure, 10 representing “a lot.” 75% reported prior ophthalmology exposure as shadowing or specific subspecialty experience. The strongest motivators to volunteer were gaining exposure to ophthalmology/current residents (67%) and volunteer experience (58%). Interpreting for Spanish-speaking patients was also identified as a motivator by 25%. Participants reported a 
variety of specialities they were interested in, but the top-reported specialty was ophthalmology (identified by 83%). Average interest in ophthalmology as a career was 6.83/10 with 0 representing no interest, 10 representing “a lot,” and 67% of participants reported an increased interest after volunteering. 100% reported that they would volunteer at this clinic again. Conclusion: On average students who volunteered at free eye clinics have some prior exposure and interest in ophthalmology and are highly motivated to volunteer in order to gain exposure to ophthalmology and volunteer experience. Majority of participants had an increased interest in ophthalmology after the clinic. This study shows that free eye clinics pique interest and provide valuable career exposure to the field of ophthalmology for medical students.

Impact of a Free Eye Clinic on Medical Students’ Exposure and Interest in Ophthalmology

BACKGROUND: Artificial Intelligence (AI) generative large language models (LLMs) are powerful and increasingly accessible tools with potential applications in healthcare education and training. The annual Orthopedic In-training Examination (OITE) is widely used to assess resident academic progress and preparation for the American Board of Orthopedic Surgery (ABOS)  Part 1.

METHODS: Open AI’s ChatGPT and Google’s Bard generative language models were administered the 2022 OITE examination. Question stems that contained  images were input without, then with a text based description of the imaging findings.

RESULTS: ChatGPT answered 69.1% of questions correctly, when provided with text describing accompanying media this increased to 77.8% correct. In contrast Bard answered 49.8% of questions correctly and this increased to 58% correct when provided with text describing imaging in question stems (p&lt;0.0001.) ChatGPT was most accurate in questions within the Shoulder category with a 90.9% accuracy. Bard performed best in the Sports subspecialty category scoring 65.4% correct.  ChatGPT performed above the published mean of Accreditation Council for Graduate Medical Education (ACGME) orthopedic resident test-takers (66%.)  
CONCLUSIONS: There is significant variability in the accuracy of publicly available AI
models on the OITE. AI generative language software may play numerous potential roles in the
future assisting orthopedic education including simulating patient presentations and clinical scenarios, customizing individual learning plans and driving evidence-based case discussion.  Further research and collaboration within the orthopedic community is required to safely adopt these tools and minimize risks associated with their use.

Performance of Two Artificial Intelligence Generative Language Models on the Orthopedic In-training Examination

Background: Fellowship Directors (FDs) and Division Chiefs (DCs) have a profound impact on the current and future trainees in orthopaedics, yet no previous study has analyzed the academic, leadership, and demographic characteristics of FDs and DCs in arthroplasty. This study’s purpose was to evaluate the demographic characteristics, scholarly work, and educational training of FDs and DCs in academic orthopaedic arthroplasty divisions in the United States.

Methods: Orthopaedic arthroplasty fellowships were identified using the American Association of Hip and Knee Surgeons Directory. Data on each FD and DC was gathered via the program’s respective websites, personal profiles, curriculum vitae, and online searches. Information collected included sex, race/ethnicity, fellowship training institution, time since fellowship graduation, academic rank, additional degrees, time at current institution, time in current role, additional leadership positions held, h-index, ResearchGate score, and ResearchGate interest score. H-index was compared between FDs and DCs and the correlation between h-index and academic rank was calculated. 

Results: There were 116 FDs and 77 DCs from 137 orthopaedic arthroplasty programs included. Of these, 20 were co-FDs and 22 were both FD and DC. Overall, 98.3% were males. For race/ethnicity, 83.8% were Caucasian, 11.0% Asian, 3.5% Hispanic, and 1.7% African American. Average age in years was 52.7, years since fellowship graduation was 18.1, years at current institution was 13.9, years in current role was 9.8. Thirty had additional degrees. Average h-index, RG, and RI scores were 16.9, 27.6, and 1309.5, respectively. Overall, 29.6% were full Professor, 26.6% Associate Professor, and 23.1% Assistant Professor. There was a moderate positive correlation between academic rank and h-index. The top fellowship programs to produce future FDs and DCs were Massachusetts General Hospital (17), Hospital for Special Surgery (17) and Mayo Clinic (13). 

Conclusion: A striking lack of diversity exists in orthopaedic arthroplasty leadership, with most FDs and DCs being white males. There was a moderate positive correlation between professorship and h-index. Furthermore, there are specific fellowship programs that play an important role in producing future arthroplasty leaders. Greater efforts are needed to increase diversity among arthroplasty leadership in United States academic orthopaedic programs.

Next from AMA Research Challenge 2023

Assessing the Accuracy and Reproducibility of Artificial Intelligence Chatbot in Answering Frequently Asked Questions Regarding Hypertension and its Associated Complications