United States

Background 
Breast cancer is one of the leading causes of cancer-related mortality worldwide, affecting millions each year. Timely detection and accurate diagnosis are essential for improving survival rates and optimizing patient outcomes. As one of the most common malignancies, breast cancer demonstrates significantly improved prognoses when identified in its early stages. This study introduces a novel machine learning model powered by artificial intelligence, developed to support healthcare professionals in the early and accurate diagnosis of breast cancer using medical imaging. By harnessing advanced computational methods, the model enhances diagnostic precision, facilitates informed clinical decision-making, and has the potential to improve patient outcomes.

Methods
A machine learning model powered by advanced AI algorithms was developed using breast imaging data to support breast cancer diagnosis. The model was trained and evaluated on a dataset comprising 3,245 patients diagnosed with breast cancer, sourced from hospitals and medical clinics across the United States. For model development, 50% of the data was randomly selected for training, while the remaining 50% was used to test and validate the model’s diagnostic performance.

Results
The developed machine learning model accurately diagnosed 90.2% of the breast cancer using the testing datasets, exceeding currently published models. 

Conclusion
Artificial intelligence in breast imaging offers a valuable tool to enhance the early and accurate diagnosis of breast cancer, supporting healthcare professionals around the world. This technology has the potential to significantly improve patient outcomes—particularly in resource-limited settings with shortages of specialized medical personnel—and to advance breast cancer care on a global scale.


2025 AMA Research Challenge – Member Premier Access

Diagnosing Breast Cancer Using Imaging and Machine Learning

poster

Congratulations to all the researchers who have qualified for the AMA Research Challenge virtual poster symposium and semifinals!

Welcome to the **largest national, multi-specialty research event** featuring research from nearly 1,000 medical students, residents, fellows, and international medical graduates across six topic areas - basic science, clinical and translational research, clinical vignettes, health systems science, medical education, and public health and health policy.

#### Access the Event
Use the buttons below or navigation menu to access the event. 

[![](https://assets.underline.io/markdown_image/1/image/d4076f2146939317a29eb9f69c53ad41.png)](https://underline.io/events/506/sessions?searchGroup=all)
[![](https://assets.underline.io/markdown_image/1/image/6bcee0b523a4874c4292ca0415d88dde.png)](https://underline.io/events/506/posters?searchGroup=all)
[![](https://assets.underline.io/markdown_image/1/image/0ca1b933deefd7731529477b45345166.png)](https://underline.io/events/506/schedule)

#### Event Details
The 2025 AMA Research Challenge includes the following symposium halls:
* **Semifinals hall:** Features research from 54 medical students, residents and international medical graduates that received top scores during first round scoring. 
* **Poster symposium hall:** Features research from nearly 1,000 medical students, residents and international medical graduates categorized by topics and subtopics.

**Here’s how to participate:**
* **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 presenters** and provide feedback to help advance their research
* **New this year: Watch live poster presentations** by topic area and engage directly with researchers

At the conclusion of this event, the AMA will announce the top five finalists who will advance to the AMA Research Challenge final round, present their research to an elite panel of judges, and compete to win a $10,000 grand prize presented by Laurel Road.

#### A message from Dr. Bobby Mukkamala, AMA President
<div style="position: relative; width: 100%; padding-bottom: 56.25%; height: 0; overflow: hidden;">
  <iframe 
    src="https://underline.io/embed/136894-ama-video?t=0"
    title="AMA Welcome Video"
    allowfullscreen
    style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border: 0;">
  </iframe>
</div>


#### Recognized Institutions
In 2025, 1,400 abstracts were submitted across six topic areas representing 174 medical schools and more than 150 residency and health care institutions. [See if your institution is represented this year!](https://drive.google.com/file/d/1eFrTqCmCeobx9pr5drJK2hllvsIpFNOY/view?usp=drive_link)

#### Grand Prize Sponsor
AMA would like to thank Laurel Road, a brand of KeyBank, for sponsoring a $10,000 grand prize for the winner of the AMA Research Challenge. Laurel Road and KeyBank are not involved in winner selection process. The winner will be announced during the AMA Research Challenge final round. Payment of the grand prize will be directly from Laurel Road/KeyBank. See the [AMA Research Challenge Official Rules for terms and conditions](https://www.ama-assn.org/system/files/research-challenge-official-rules.pdf).

[![](https://assets.underline.io/uploads/markdown_image/1/image/611e331d4a0eba0920f240cfc04b4d77.png)](https://www.laurelroad.com/)

#### Code of Conduct
<html>
AMA has a code of conduct for its meetings – we ask all attendees to adhere to this code to participate in this year’s meeting. Please read and <a href="https://www.ama-assn.org/about/code-conduct" target="_blank">learn more here</a> about your conduct requirements.
<br><br>
	</html>
	

#### Semifinals Judges

You need to log in with the email address you registered with. Access credentials have been sent to your email. 

Please be sure to check your spam and other email folders if you do not see an email confirmation right away.

Need help? Contact us at <support@underline.io>.

Please log in to explore this event.

If you have used the AMA Research Challenge platform previously, please click the ‘Log-in’ button below. If you do not remember your password, please click “forgot password” and follow prompts to reset your password. After resetting your password, please refresh your webpage to access the platform. 

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/member-benefits/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 and score posters in the semifinals (an AMA member exclusive opportunity).

Background: The management of obstetric patients with concurrent hypercoagulable disorders, hypertensive emergencies, and cerebrovascular complications presents significant anesthetic challenges. Factor V Leiden deficiency, the most common inherited thrombophilia, significantly increases the risk of venous thromboembolism (VTE) in pregnancy, already a hypercoagulable state [1]. Moreover, preeclampsia with severe features, associated with end-organ damage, is a leading cause of maternal morbidity and mortality, contributing to complications such as intracranial hemorrhage (ICH) [2]. These in addition to a premature gestation complicated by posterior placenta previa, requires a multidisciplinary and individualized anesthetic approach.

Case: A 33-year-old G5P0141 at 27 weeks and 4 days gestation presented to the ED with a new-onset headache and nausea following progressively worsening blood pressures (peaking at 210/130) and lower extremity edema. She had no prior history of hypertension. Her medical history was significant for Factor V Leiden deficiency with a history of DVT/PE while on oral contraceptives, pulmonary hypertension. CT imaging revealed a 2 cm right posterior parietal/superior occipital intraparenchymal hemorrhage (IPH) and a small-volume subarachnoid hemorrhage (SAH) without evidence of mass effect or midline shift. MRI findings were consistent with posterior reversible encephalopathy syndrome (PRES) and hypertensive-mediated hemorrhage. The patient was transferred to the neuro-ICU for close monitoring. She received magnesium sulfate and levetiracetam for seizure prophylaxis, betamethasone for fetal lung maturation, nicardipine infusion for blood pressure control and her anticoagulation was held. Given the maternal and fetal risks, a multidisciplinary team (obstetrics, neurology) recommended early delivery via Cesarean section.

Anesthetic Management
The patient underwent a primary low transverse Cesarean section (LTCS) under a combined spinal-epidural (CSE) technique. Neuraxial anesthesia was chosen over general anesthesia to mitigate risks of airway complications, ICP fluctuations, and hemodynamic instability. She arrived to the OR neurologically intact with an infusion of nicardipine at 10mg/hr with initial BP readings at 112/56. In the seated position, a 17G Tuohy needle was used to access the epidural space with a loss of resistance syringe. A 25G pencil-tip spinal needle was inserted at L3-L4, delivering 0.75% bupivacaine (12 mg) intrathecal. Finally, a 19G epidural catheter was placed. Her highest blood pressure throughout the placement was 137/64 at the end of the procedure. Her highest blood pressure was during surgical prep 160/96 though trended down throughout the rest of the case leading to a reduction and then discontinuation of nicardipine. A live female infant was delivered in cephalic presentation with APGAR scores of 6, 7, and 8. Due to uterine atony, the patient received high-dose oxytocin and 250mcg of Carboprost IM, with satisfactory uterine tone achieved. Her intraoperative course was complicated with nausea and mild surgical discomfort both treated with IV medications. She also received 3mg morphine intra-epidural for postoperative pain. Estimated blood loss (EBL) was 500 mL, and the patient was transferred back to the ICU in stable condition infusing oxytocin.

Postoperative Course
The epidural was removed 3 hours after the procedure. Postoperative pain control consisted only of acetaminophen as needed. The patient required as needed IV antihypertensives starting 12 hours after the procedure. On POD1 repeat imaging did not reveal notable change in either intracranial hemorrhage, and a heparin infusion was restarted. On POD2 her highest post op blood pressure was 181/93 and the nicardipine infusion was restarted, On POD3 she successfully switched the oral and as needed IV antihypertensives. Imaging on POD2 and 3 of the intracranial hemorrhages were also stable. On POD 5 the patient was transitioned to enoxaparin twice daily and discharged on that regimen on POD7. Outpatient follow-up includes a 1-week postop visit with Obstetrics and repeat neurologic imaging 3 months after discharge.

Discussion: This case demonstrates the complexities of managing anesthesia in high-risk obstetric patients with concurrent hypercoagulability, hypertensive disorders, and intracranial pathology. Pregnancy increases both cardiovascular demands and coagulation activation, and Factor V Leiden deficiency further exacerbates VTE risks. However, intracranial hemorrhage raises concerns for anticoagulation and neuraxial anesthesia due to the potential risk of elevated intracranial pressure and epidural hematoma [3]. Neuraxial anesthesia was carefully chosen in this case due to its benefits in maintaining hemodynamic stability, avoiding ICP fluctuations, and optimizing postoperative pain control. In patients with intracranial hemorrhage, careful BP management is essential to alter a stable intracranial bleed. Postoperatively, the use of multimodal analgesia ensured effective pain control while minimizing opioid requirements, thus reducing the risk of respiratory depression and prolonged immobility. This case underscores the importance of multidisciplinary collaboration and individualized anesthetic planning to ensure maternal and fetal safety. In this instance, the neuraxial anesthetic, not without risk, provided satisfactory hemodynamic management, pain control, and was without neurologic complications.

Neuraxial Anesthesia for Cesarean Delivery in a Patient with Intracranial Hemorrhage and Chronic Anticoagulation

Background
Postpartum hemorrhage (PPH) is a leading cause of maternal morbidity and mortality worldwide and often necessitates blood transfusion. However, Jehovah’s Witness patients may decline transfusions due to religious beliefs, creating significant challenges in emergent obstetric care. Respecting patient autonomy while striving to preserve life requires innovative, multidisciplinary strategies and non-blood approaches to hemorrhage control and anemia management.
Case Presentation
A 28-year-old G1P0 woman at 40 weeks gestation, with no significant past medical history, presented for induction of labor. After two days, she progressed to 5 cm dilation but developed a category II fetal heart tracing. She subsequently elected to undergo a primary cesarean section. As a devout Jehovah’s Witness, she firmly refused all blood products and human-derived albumin but consented to intraoperative cell salvage.
Her preoperative hemoglobin was 12.1 g/dL. During the cesarean section, her estimated blood loss was 800 mL, with approximately 100 mL of autologous blood returned via cell salvage. Postoperatively, her hemoglobin decreased to 8.6 g/dL. She then developed significant PPH due to uterine atony. Despite aggressive management—including uterotonics (oxytocin, methylergometrine, misoprostol), tranexamic acid, and placement of a Jada balloon—she continued to have brisk bleeding.
She underwent a supracervical hysterectomy, which controlled the hemorrhage but resulted in an additional estimated blood loss of 1 liter. Postoperatively, her hemoglobin fell to 2.8 g/dL, with a nadir of 1.5 g/dL approximately ten days later. Her postoperative course was complicated by hemorrhagic shock, deep vein thrombosis, pulmonary embolism, hospital-acquired pneumonia, upper gastrointestinal bleeding, and ascites.
She was managed without transfusion, receiving non-blood interventions including intravenous iron, vitamin B12, folic acid, epoetin alfa, and albumin. She was transferred to a specialized bloodless medicine facility when her hemoglobin rose to 4.6 g/dL, where she continued treatment with Retacrit, Venofer, and supportive care. She was ultimately discharged home 44 days postpartum with a hemoglobin of 7.4 g/dL.
Discussion
This case highlights the ethical and clinical complexities of managing severe postpartum hemorrhage in a patient declining blood transfusion. Respecting patient autonomy, even when refusal may be life-threatening, is paramount. Effective non-blood strategies for managing anemia in such cases include cell salvage, tranexamic acid, erythropoiesis-stimulating agents, intravenous iron, vitamin supplementation, meticulous surgical techniques, and minimizing iatrogenic blood loss. Early multidisciplinary involvement and proactive planning are essential to optimize outcomes while honoring patient beliefs. Survival is possible despite profound anemia, underscoring the importance of patient-centered, innovative care in obstetric emergencies.

Management of Severe Postpartum Hemorrhage in a Jehovah’s Witness Patient: Balancing Life-Saving Care and Patient Autonomy

Background 
Magnet ingestion in the pediatric population has significantly increased over the past decade and is associated with high levels of morbidity and mortality. The strong attraction between two ingested magnets can cause connection through intestinal walls, leading to ischemia at the connection site, necrosis, and perforation/fistula formation. The clinical presentation in young or non-verbal children can be very non-specific and lead to a delay in diagnosis. 

Case Presentation 
Patient is an 18-month-old male with no past medical history who presented three times to the emergency department (ED) for fussiness and abdominal pain. On initial presentation, his father reported repeated episodes of “curling-up” and crying with no bowel movement for three days. The medical team recommended transfer to a pediatric ED. At this visit the next day, a new concern for fever was noted. The patient was given an enema and passed an oral intake challenge. He was discharged and prescribed polyethylene glycol for constipation. Two days after discharge, the patient re-presented to the ED after continued decreased intake, fussiness, constipation, and a new episode of vomiting. 

On exam, the abdomen was noted to be tender to palpation with distension noted, no guarding/rebound. 

A complete metabolic panel, complete blood count, and inflammatory markers were obtained without significant abnormalities. A respiratory viral panel was positive for rhinovirus/enterovirus. Urinalysis showed no evidence of urinary tract infection. An ultrasound of the appendix and ileocolic region was within normal limits. Abdominal X-ray performed at the third ED presentation showed concern for ingestion of a foreign body, possibly a metallic bracelet. 

The patient was admitted and X-ray was repeated 6 hours later with no changes. Due to the failure to advance, configuration of the object, and metallic appearance, suspicion for magnet ingestion was greatly increased. The surgery team was consulted and took the patient for urgent exploratory laparotomy. A complex entero-enteric fistulae was visualized that had created an internal hernia, compressing a loop of the right colon. This was repaired and the magnets removed in one string. The patient remained admitted for 7 days for recovery. 

Discussion 
This case highlights the dangers of magnet ingestion and the importance of considering this differential in young children with abdominal pain as the clinical presentation can be very non-specific. Seeing a “ring” or “bracelet” sign of metallic objects is also an important clinical indication of possible magnet ingestion that should be further investigated and recognized.

Fussiness and abdominal pain in an 18-month-old male, a surprising connection

Title
Osteopathic Diagnoses and Treatments Among United States Trauma Patients

Authors:
Hanna Harris
Aviya Distefano
Kaitlin M Bowers, DO
Dhimitri A Nikolla, DO, MS

Affiliations:
Department of Emergency Medicine, Jerry M. Wallace School of Osteopathic Medicine, Campbell University, Lillington, USA.
Department of Emergency Medicine, Jerry M. Wallace School of Osteopathic Medicine, Campbell University, Lillington, USA.
Department of Emergency Medicine, Jerry M. Wallace School of Osteopathic Medicine, Campbell University, Lillington, USA.
Department of Emergency Medicine, Allegheny Health Network – Saint Vincent Hospital, Erie, PA, USA.

Background
Osteopathic manipulative medicine (OMM) encompasses a wide range of diagnostic and treatment techniques for acute and chronic conditions. However, few studies have evaluated the use of OMM on acutely injured patients, and none have done so on a nationwide scale. 

Objectives
We aimed to estimate the incidence of OMM use among trauma patients in United States (US) emergency departments (ED) and provide demographic and clinical characteristics of these osteopathic trauma encounters. 

Methods
Design and Setting
We performed a retrospective, secondary analysis of the 2018 to 2021 Nationwide Emergency Department Sample (NEDS) datasets, Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality. Data are from encounters from hospital-owned US EDs. The Allegheny Health Network Institutional Review Board deemed the study exempt.

Patients and Interventions
We included all ED encounters with a primary or secondary injury diagnosis identified by the NEDS I10_INJURY variable and an osteopathic diagnosis or treatment. Osteopathic diagnoses were identified by the International Classification of Diseases-10 (ICD-10) Clinical Modification (CM) codes M9900-M9909. Osteopathic treatments were identified by either ICD-10 Procedure Coding System (PCS) codes 7W00X0Z-7W09X9Z or Current Procedural Terminology / Healthcare Common Procedure Coding System (CPT/HCPCS) codes 98925-98929.

Main Outcome Measures
We used the svyciprop function with the beta method from the survey package in R to calculate the weighted incidence of osteopathic trauma encounters per 100,000 trauma patients accounting for complex survey design. We provided unweighted descriptive statistics for demographic and clinical characteristics. 

Results
Among over 24 million trauma encounters, we identified 1,356 with an osteopathic diagnosis or treatment. The weighted incidence of osteopathic encounters was 6.0 (95% CI 4.0 to 8.5) per 100k visits. The median patient age was 50 years (IQR 32 to 66) and 753 (55.5%) were female. 459 (33.8%) encounters in level one or two trauma centers. 359 (26.5%) had injuries due to a fall and 536 (39.5%) were admitted. 852 (62.8%) had an ISS of ≤1 with a median score of 1 (IQR 1 to 4) and a maximum score of 50. The most common ICD-10-CM code was segmental and somatic dysfunction of thoracic region (485 of 2,877 codes, 16.9%), and 553 of 1,052 (52.6%) of ICD-10-PCS codes were for fascial release. 

Conclusion
Osteopathic encounters were uncommon among acutely injured patients in US EDs. Most had minor injuries, but over a third were admitted. Additional research is needed to describe the indications for OMM among acutely injured patients, particularly if OMM is used to treat acute injury extrinsic to these locations.

Osteopathic Diagnoses and Treatments Amongst United States Trauma Patients

Background
Ultrasound technology has been utilized in diagnostic criteria since the early 1950s. With its minimal risk of complications and clear visualization, it is a first line diagnostic tool for many specialties. Some barriers to ultrasound in the past are lack of access and lack of training. The Butterfly Ultrasound is a hand-held, portable ultrasound that comes with user-based modules. With these modules and portability of technology, the goal of limiting barriers is more feasible.
Methods
Four OMS-II students attended a week-long medical outreach trip in the Dominican Republic where each were assigned a Butterfly ultrasound device with access to online educational modules. A list of modules within the Butterfly platform was compiled prior to the mission trip, which served as a guide to enhance students' understanding when learning to operate the device. Students had no previous ultrasound experience prior to this project and relied solely on these educational modules. During the international mission trip, students performed ultrasounds on patients when clinically indicated. Data was collected to assess the primary outcome measures, which included the number of patients scanned, the type of ultrasound performed, the reason for the ultrasound, and the health outcome. Improved health outcomes were based on the following criteria: if the patient was diagnosed with a condition and/or referred for further follow-up, if the visit was for a general wellness visit for prenatal care, or if a condition was ruled out and the patient was referred for additional follow-up. The data was summarized using a descriptive statistical analysis.
Results
505 patients were seen in local villages, with another estimated 80-100 patients seen in the local clinics. Students performed 74 ultrasound scans on patients with clinical indications to aid in their diagnosis/plan of care, which was about 12.2-12.6% of the total patients that were seen. Students utilized renal, cardiac, gallbladder, OB/GYN wellness, spleen, appendix, lungs, and soft tissue ultrasounds, where 64.1% of scans were for OB/GYN purposes with the rest being relatively equivalent. Of the patients that received ultrasounds, 87.3% had associated improved health outcomes, while 12.7% had no relative change in health outcomes. 
Conclusion
In a mission trip environment where medical resources were lacking, the Butterfly ultrasound provided a method for more comprehensive evaluation of patients. The modules used for education provided a platform of experience that can now be expanded upon in the future to improve patient care and further evaluate the role the Butterfly ultrasound has on patient outcomes.


EVALUATING THE FEASIBILITY AND EFFICACY OF USING MODULE BASED LEARNING IN ULTRASOUND-NAIVE MEDICAL STUDENTS TO OPTIMIZE HEALTH OUTCOMES IN THE SETTING OF INTERNATIONAL MEDICAL OUTREACH 

Background:
Lipoprotein(a) [Lp(a)] is a genetically inherited low-density lipoprotein-like particle that is an independent and significant risk-enhancer for atherosclerotic cardiovascular disease (ASCVD). Despite Class IIa recommendations by the ACC/AHA and ESC for one-time screening, provider ordering rates remain below 1% across many U.S. health systems. Our institution had no structured educational programming related to Lp(a), prompting a quality improvement (QI) initiative focused on increasing awareness and screening intent.
Objective:
To increase provider-reported intention to screen for Lp(a) by at least 30% among internal medicine residents and cardiology/primary care providers through a brief, targeted educational intervention.
Methods:
A multidisciplinary QI team developed a 45-minute interactive session combining a 30-minute didactic lecture with 15 minutes of discussion. The session covered Lp(a)'s clinical relevance, guideline recommendations, and appropriate use cases. It was delivered separately to three provider cohorts: internal medicine residents, primary care physicians, and cardiologists. Trifold handouts and clinic posters reinforced key points. Anonymous pre- and post-session surveys measured self-reported knowledge, guideline familiarity, and intention to screen. Unique participant-generated codes allowed matched response analysis without identifying data. Barriers to screening (e.g., knowledge gaps, EMR workflow, insurance coverage) were also assessed.
Results:
Preliminary data from approximately 40 providers showed >80% attendance and active participation. Early post-survey responses indicated improved familiarity with Lp(a) guidelines and increased confidence in identifying appropriate candidates for screening. While final survey data analysis is pending, we anticipate a >=30% rise in screening intent based on previous similar QI interventions. Qualitative responses suggested reduced uncertainty around guideline criteria and improved recognition of Lp(a) as a valuable risk stratification tool.
Conclusion:
A focused, low-resource educational session significantly improved provider engagement with Lp(a) screening guidelines and showed early signs of shifting clinical intent. This model could be integrated into annual onboarding and CME sessions and adapted to support guideline adoption for other underutilized ASCVD risk markers. Broader application may enhance preventive cardiology practices and support health system alignment with national care standards.

Improving Provider Awareness and Screening Practices for Elevated Lipoprotein(a) in the Outpatient Setting

Background
Due to its diagnostic complexity, Selective Mutism (SM) is underrecognized and misdiagnosed in up to 70% of children, often as social anxiety disorders and/or autism. Accurate diagnosis includes family and clinician input in multiple settings. Families rely on patient education materials (PEMs) to inform symptomatology and treatment, which frequently are written above the recommended 6th-8th grade reading level. Prior studies show that large language models (LLMs) can significantly improve PEM readability while preserving clinical accuracy. This study aims to assess whether tools like ChatGPT-4.5 and Claude Sonnet 4 can serve as an innovative solution to simplifying critical public health information for SM.

Methods
We conducted a Google search for “What is selective mutism?” and selected the first 20 online PEMs. Text was extracted and analyzed for readability using Readability.com, which provided Flesch-Kincaid Grade Level (FKGL) and Flesch Reading Ease (FRE) scores. PEMs were rewritten by ChatGPT-4.5 and Claude Sonnet 4 using the prompt: “Read this carefully. Then rewrite it clearly so a 6th grader could understand it.” Readability was reassessed, and outputs were evaluated for clinical alignment with DSM-5 criteria. Researchers independently scored each LLM output. Inter-rater reliability was assessed via StatsToDo using a two-way random effects model for absolute agreement (ICC[2,1]). Paired t-tests (GraphPad Prism v.10.5.0) compared DSM-5 alignment scores between original and LLM-enhanced PEMs.

Results
Selective Mutism PEMs reached readability scores of a post-high school reading level. Google PEMs yielded a mean FKGL score of 10.23 and FRE score of 47.35. After enhancement with ChatGPT-4.5, the mean FKGL score was 7.57 and mean FRE score was 61.29. Claude Sonnet 4 yielded a mean FKGL score of 6.08 with a mean FRE score of 71.2. Inter-rater reliability was high across all sources: 0.96 for Google (95% CI: 0.90-0.98), 0.91 for ChatGPT-4.5 (95% CI: 0.77-0.96), and 0.84 for Claude Sonnet 4 (95% CI: 0.68-0.94). Paired t-test assessed clinical accuracy. No significant difference was found between Google and ChatGPT-4.5 (p=0.45); however, Claude Sonnet 4 differed significantly from Google (p=0.01) and ChatGPT-4.5 (p=0.003).

Conclusion
LLMs, like ChatGPT-4.5 and Claude Sonnet 4, improved the readability of PEMs on SM. ChatGPT-4.5 slightly outperformed Claude Sonnet 4 in preserving key DSM-5 diagnostic features, while Claude achieved greater gains in readability. Findings suggest that LLMs are effective tools for generating clearer, clinically aligned PEMs on underrecognized psychiatric conditions; thus, supporting early identification, intervention, and public health communication.

Can LLMs, like ChatGPT-4.5 and Claude Sonnet 4, Make Selective Mutism Patient Education Materials (PEMs) More Accessible?

The opioid epidemic remains a critical public health crisis, with thousands of preventable deaths occurring annually. Naloxone, an opioid antagonist, can rapidly reverse overdoses, but effective use often depends on proper training. This study investigates how Naloxone training impacts a responder’s confidence, preparedness, and effectiveness during overdose situations. Participants-including first responders, bystanders, and family members-received Naloxone kits with QR codes linking to follow-up surveys after real-world administrations. Surveys assessed comfort level, response actions, and perceived preparedness, and results were compared between trained and untrained responders using descriptive statistics and t-tests. Preliminary findings show that training significantly increases confidence, reduces hesitation, and improves response accuracy. Participants reported feeling better prepared and more willing to act, often recognizing overdose signs they might have otherwise missed. These results highlight the importance of accessible community-based Naloxone training in reducing overdose mortality and strengthening emergency response efforts.


The Impact of Naloxone Trainings on Opioid Overdoses

Abstract Title: Disparities in Cirrhosis Mortality in Tennessee: A 17-Year Analysis of Policy, Socioeconomic, and Geographic Determinants

Background: 
Cirrhosis mortality rates in Tennessee (TN) have surged by 69% (2005–2022), far exceeding neighboring states (KY, AL, GA) despite comparable demographics. This study investigates the role of alcohol policies, healthcare access, and socioeconomic factors in driving TN’s divergent trends.

Methods
We analyzed CDC mortality data (2005–2022) for TN and peer states (KY, AL, GA), integrating state-level policies (Medicaid expansion, alcohol regulations) and county-level socioeconomic indicators (poverty, uninsured rates). Trends were visualized using segmented regression and geospatial mapping.

Results:
1. TN’s cirrhosis mortality rate (16.6/100,000 in 2022) surpassed KY (16.0) and AL (15.5), while GA (12.6) maintained the lowest rate.
2. Policy gaps: Non-expansion of Medicaid in TN correlated with steeper mortality increases post-2014 vs. Medicaid-expanded KY.
3. Socioeconomic drivers: Rural TN counties with poverty rates >20% had mortality rates 2.3× higher than urban counties.


Conclusion:
TN’s rising cirrhosis mortality reflects systemic gaps in addiction treatment access and alcohol policy enforcement. Prioritizing Medicaid expansion, rural harm reduction programs, and stricter alcohol sales regulations could mitigate disparities.
Implications for Policy:
• Advocate for Medicaid expansion to improve treatment access.
• Target high-risk rural counties with mobile clinics and screening programs.
• Benchmark against GA’s successful urban healthcare models.


Disparities in Cirrhosis Mortality in Tennessee: A 17 year Analysis of Social Determinants, Policy and Demographics

Background: Disparities in healthcare outcomes are increasingly recognized across surgical disciplines, including cardiac surgery. This study assesses inequities in predicted postoperative outcomes for isolated coronary artery bypass grafting (CABG) surgery using the Society of Thoracic Surgeons (STS) Short-term/Operative Risk Calculator, with a focus on how sex, race, and insurance status influence risk projections.
Methods: Predicted outcomes were modeled using a standardized ‘control’ patient profile: a 60-year-old individual undergoing their first cardiovascular surgery, classified as urgent, with three-vessel coronary artery disease and no comorbidities. Variables analyzed included patient sex, race, and insurance type (Medicaid vs. commercial insurance). Two primary STS risk metrics were examined: (1) combined morbidity and mortality, and (2) probability of prolonged hospital stay (defined as greater than 14 days).
Results: Across all demographic subgroups, patients with Medicaid coverage faced consistently higher projected risks of morbidity, mortality, and extended hospital stays compared to those with commercial insurance. Female patients showed higher predicted morbidity and mortality risks than males of the same race, regardless of insurance status. The subgroup with the highest predicted risk included Asian and Black/African American females with Medicaid insurance.
Conclusion: These findings highlight significant disparities in predicted CABG outcomes based on intersecting demographic and socioeconomic factors. Medicaid coverage and female sex are associated with increased surgical risk, potentially due to systemic and institutional drivers. The data suggest a need for further investigation into the structural causes of these inequities and call for equity-centered reforms in preoperative risk assessment and healthcare delivery.

Next from 2025 AMA Research Challenge – Member Premier Access

Neuraxial Anesthesia for Cesarean Delivery in a Patient with Intracranial Hemorrhage and Chronic Anticoagulation