United States

### Background
Racial disparities in Alzheimer’s disease and related dementias (ADRD) are well-documented, yet the contribution of neighborhood socioeconomic context to these disparities among breast cancer survivors remains unclear. The Index of Concentration at the Extremes (ICE) captures neighborhood-level deprivation and affluence, potentially mediating racial differences in ADRD diagnosis. Understanding the role of neighborhood-level factors may explain mechanisms behind these disparities.

### Objective
To assess whether neighborhood-level socioeconomic conditions, measured by ICE, mediate the racial disparity in ADRD diagnosis among older female breast cancer survivors.

### Methods
We analyzed SEER-Medicare linked data from 17,741 women aged 65 and older diagnosed with breast cancer (2007–2009) with up to 10 years of follow-up. ADRD diagnosis was identified through diagnostic codes. We conducted causal mediation analysis to estimate the average causal mediation effect (ACME), total effect, and proportion mediated by ICE on the association between race (Black vs. White) and ADRD diagnosis, adjusting for age, comorbidities using the Charlson Comorbidity Index (CCI), treatment type, and neighborhood -level poverty indicator.

### Results
Race was significantly associated with ADRD diagnosis (total effect estimate = 0.052, 95% CI 0.028–0.077, p &lt; 0.001). ICE significantly mediated this association, with an ACME of 0.049 (95% CI 0.035–0.063, p &lt; 0.001), explaining approximately 93% of the total effect. This strongly suggests that neighborhood-level deprivation/advantage (ICE) explains most of the racial disparity in ADRD diagnoses in the dataset, after adjusting for age, neighborhood-level poverty indicator, treatment type, and CCI. The mediation findings remained robust in sensitivity analyses adjusting for additional covariates.

### Conclusion
Neighborhood-level deprivation, as captured by ICE, substantially mediates racial disparities in ADRD diagnosis among breast cancer survivors, highlighting the critical role of social
determinants and structural inequities on cognitive health outcomes. These findings suggest that beyond biological factors, such as hormone-modulating therapy-related cognitive effects, social context profoundly shapes racial disparities in cognitive health outcomes. This study shows the importance of neighborhood socioeconomic context in understanding racial disparities in ADRD diagnosis among breast cancer survivors. Further analyses incorporating additional covariates and stratified mediation will clarify the pathways influencing these disparities. Targeted interventions addressing neighborhood disadvantage may be essential for advancing equity in dementia prevention and care.

AMA Research Challenge 2025 

Neighborhood Socioeconomic Status and Racial Disparities in ADRD in Breast Cancer Survivors

**ICWSM-2022 will be held from June 6th – 9th in a hybrid format, both in-person in Atlanta, Georgia and online.**

**The Workshop and Tutorial Sessions are accessible to those that registered separately for these sessions. If you registered to attend but are not able to access the these sessions from the [Schedule tab](https://underline.io/events/291/schedule?day=2022-06-06T04%3A00%3A00.000Z), please email icwsm@underline.io
**

The International AAAI Conference on Web and Social Media (ICWSM) is a forum for researchers from multiple disciplines to come together to share knowledge, discuss ideas, exchange information, and learn about cutting-edge research in diverse fields with the common theme of online social media. This overall theme includes research in new perspectives in social theories, as well as computational algorithms for analyzing social media. ICWSM is a singularly fitting venue for research that blends social science and computational approaches to answer important and challenging questions about human social behavior through social media while advancing computational tools for vast and unstructured data.


ICWSM, now in its sixteenth year, has become one of the premier venues for computational social science, and previous years of ICWSM have featured papers, posters, and demos that draw upon network science, machine learning, computational linguistics, sociology, communication, and political science. The uniqueness of the venue and the quality of submissions have contributed to a rapidly growing conference, and a competitive acceptance rate of approximately 20% for full-length research papers published in the proceedings by the Association for the Advancement of Artificial Intelligence (AAAI).


**ICWSM Code of Conduct**

All persons, organizations and entities that attend AAAI conferences and events are subject to the standards of conduct set forth on the [AAAI Code of Conduct for Events and Conferences](https://aaai.org/Conferences/code-of-conduct.php). AAAI expects all community members to formally endorse this code of conduct, and to actively prevent and discourage any undesired behaviors. Everyone should feel empowered to politely engage when they or others are disrespected, and to raise awareness and understanding of this code of conduct. AAAI event participants asked to stop their unacceptable behavior are expected to comply immediately. Sponsors are also subject to this code of conduct in their participation in AAAI events.


Additionally, participants are encouraged to be courteous when sharing screen captures and photographs of conference events. Seek permission when possible and respect requests to take down images if those featured ask. Concerns around code of conduct or inclusion may be sent to icwsm22@aaai.org. If you have any concerns or items to report, please reach out to the General Chairs: Diyi Yang and Yelena Mejova.<br><br>


A ticket is required to attend the event, please use the below URL to register for the event if you have not already done so:

https://aaaiconf.cventevents.com/event/be8c26b0-a76f-4753-babb-de2bd9d2a619/summary

Virtual-Only:
$200 AAAI Member
$125 AAAI Student Member
$225 Nonmember
$150 Nonmember Student

Virtual Silver: Includes discounted conference registration, plus a one-year online new or renewed membership in AAAI.
$299 Silver Regular
$174 Silver Student

Virtual - Workshop and Tutorials:
$45 Regular
$25 Student


Registration Is Required

ICWSM 2022

The International AAAI Conference on Web and Social Media (ICWSM) is a forum for researchers from multiple disciplines to come together to share knowledge, discuss ideas, exchange information, and learn about cutting-edge research in diverse fields with the common theme of online social media. This overall theme includes research in new perspectives in social theories, as well as computational algorithms for analyzing social media. ICWSM is a singularly fitting venue for research that blends social science and computational approaches to answer important and challenging questions about human social behavior through social media while advancing computational tools for vast and unstructured data.

### Background
Racial disparities in Alzheimer’s disease and related dementias (ADRD) are well-documented, yet the contribution of neighborhood socioeconomic context to these disparities among breast cancer survivors remains unclear. The Index of Concentration at the Extremes (ICE) captures neighborhood-level deprivation and affluence, potentially mediating racial differences in ADRD diagnosis. Understanding the role of neighborhood-level factors may explain mechanisms behind these disparities.

### Objective
To assess whether neighborhood-level socioeconomic conditions, measured by ICE, mediate the racial disparity in ADRD diagnosis among older female breast cancer survivors.

### Methods
We analyzed SEER-Medicare linked data from 17,741 women aged 65 and older diagnosed with breast cancer (2007–2009) with up to 10 years of follow-up. ADRD diagnosis was identified through diagnostic codes. We conducted causal mediation analysis to estimate the average causal mediation effect (ACME), total effect, and proportion mediated by ICE on the association between race (Black vs. White) and ADRD diagnosis, adjusting for age, comorbidities using the Charlson Comorbidity Index (CCI), treatment type, and neighborhood -level poverty indicator.

### Results
Race was significantly associated with ADRD diagnosis (total effect estimate = 0.052, 95% CI 0.028–0.077, p < 0.001). ICE significantly mediated this association, with an ACME of 0.049 (95% CI 0.035–0.063, p < 0.001), explaining approximately 93% of the total effect. This strongly suggests that neighborhood-level deprivation/advantage (ICE) explains most of the racial disparity in ADRD diagnoses in the dataset, after adjusting for age, neighborhood-level poverty indicator, treatment type, and CCI. The mediation findings remained robust in sensitivity analyses adjusting for additional covariates.

### Conclusion
Neighborhood-level deprivation, as captured by ICE, substantially mediates racial disparities in ADRD diagnosis among breast cancer survivors, highlighting the critical role of social
determinants and structural inequities on cognitive health outcomes. These findings suggest that beyond biological factors, such as hormone-modulating therapy-related cognitive effects, social context profoundly shapes racial disparities in cognitive health outcomes. This study shows the importance of neighborhood socioeconomic context in understanding racial disparities in ADRD diagnosis among breast cancer survivors. Further analyses incorporating additional covariates and stratified mediation will clarify the pathways influencing these disparities. Targeted interventions addressing neighborhood disadvantage may be essential for advancing equity in dementia prevention and care.

technical paper

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)

#### 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 announced the top five finalists who advanced to the AMA Research Challenge final round, presented their research to an elite panel of judges, and competed to win a $10,000 grand prize presented by Laurel Road.

**Addison Shenk, a fourth-year medical student from Edward Via Virginia College of Osteopathic Medicine, was the winner of the 2025 AMA Research Challenge and was awarded the grand prize of $10,000, 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. **

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 
Aging confers the most significant risk for dementia, yet the underlying mechanisms remain poorly understood. While the amyloid cascade hypothesis has dominated the field for decades, a large proportion of variance in cognitive impairment remains unexplained by classical proteinopathies. Thromboinflammation—a vascular and immune signaling axis involving brain barrier dysfunction and protein extravasation—may represent an upstream initiator of neuronal vulnerability that precedes amyloid pathology. Recent work by our colleague shows fibrinogen, a peripheral coagulation protein, to exhibit one of the earliest age-related increases in cerebrospinal fluid, yet its role as a driver of brain degeneration in living humans remains unexplored. 

### Methods 
We present the most comprehensive investigation of thromboinflammation as a driver of degenerative brain changes in aging living humans. We integrated high-dimensional cerebrospinal fluid proteomics (~7,000 proteins) across 1,655 individuals from three independent cohorts with single-cell transcriptomics, brain imaging, and cognitive data. Using CSF fibrinogen levels as a quantifiable marker of brain barrier dysfunction, we employed systematic mediation analysis to identify molecular pathways linking vascular compromise to synaptic vulnerability in amyloid-negative, cognitively typical older adults, with validation across independent cohorts and testing in amyloid-positive states. 

### Results
We identified a reproducible molecular network linking brain barrier dysfunction to white matter degeneration, synaptic loss, and cognitive slowing—well before amyloid pathology appears. Fibrinogen emerged as a central hub of brain aging biology, with higher CSF levels significantly associating with elevated markers of synaptic dysfunction (pTau181, neurogranin, GAP43, SNAP25) and cognitive slowing in amyloid-negative adults. We identified 53 validated molecular mediators of fibrinogen's neurotoxic effects, enriched in neurovascular unit cells and demonstrating coordinated dysregulation of angiogenesis, fibrosis, and immune signaling. These mediator signatures correlated with compromised white matter integrity in tracts critical for cognitive processing speed. Critically, fibrinogen associated with tauopathy independent of amyloid buildup, and these associations persisted robustly in amyloid-positive individuals while remaining independent of amyloid burden, demonstrating relevance across the neurodegeneration spectrum. 

### Conclusion 
Our findings offer a paradigm-shifting answer to why aging confers powerful dementia risk by positioning thromboinflammation as an upstream initiator of neuronal vulnerability that challenges the prevailing temporal hierarchy of neurodegeneration. This work reframes vascular disease from comorbidity to causality, introducing a vascular-centered model of brain aging that bypasses classical proteinopathy models. The identified molecular mediators provide an upstream, druggable axis with immediate implications for biomarker development, clinical trial stratification, and brain healthspan-preserving interventions across the aging spectrum. "

A proteomic signature of vascular disease is associated with the aging brain's vulnerability to degeneration preceding accumulation of amyloid pathology

### Background
Lung cancer is the leading cause of cancer-related mortality in the United States, with outcomes disproportionately worse among uninsured and socioeconomically disadvantaged populations. The Affordable Care Act (ACA), through Medicaid expansion, sought to improve access to care, yet its long-term impact on lung cancer survival remains incompletely defined. 

### Objective
To evaluate the association between Medicaid expansion and overall mortality in patients with lung cancer, and to determine whether effects varied by stage at diagnosis, race/ethnicity, socioeconomic status, and treatment modality. 

### Methods
We conducted a retrospective cohort study using the National Cancer Database, including patients aged 40–64 years diagnosed with lung cancer between 2006 and 2021. States were categorized as early Medicaid expansion (by January 2014) or non-expansion. The study period was divided into pre-expansion (2006–2013) and post-expansion (2014–2021). A Difference-in-Differences (DiD) analytic approach within a Cox proportional hazards framework was used to estimate the effect of Medicaid expansion on all-cause mortality, adjusting for demographic, clinical, and socioeconomic factors. 

### Results
The study cohort comprised 369,251 patients, predominantly White (74.7%) and male (52.7%), with a mean age of 56.8 ± 5.6 years. A total of 165,364 individuals (44.8%) resided in early Medicaid expansion states. Medicaid expansion was associated with a 45.2% reduction in overall mortality. Stratified analyses demonstrated reductions in the hazard of overall mortality across all racial and ethnic groups, with the greatest benefit observed among Hispanic patients (DID: –0.506; 95% CI, –0.729 to –0.283; p < 0.001). Expansion states also showed increased early-stage diagnoses, reduced uninsured rates, and higher surgical utilization following the ACA. 

### Conclusions
Medicaid expansion was significantly associated with improved lung cancer survival across demographic and clinical subgroups. These findings underscore the role of insurance policy in shaping oncologic outcomes and suggest that Medicaid expansion may serve as a critical strategy to enhance access to timely diagnosis and treatment. 



Medicaid Expansion and Overall Mortality among Individuals Living with Lung Cancer

### Background & Objectives
The 2025 ACC/AHA hypertension guidelines incorporate the Predicting Risk of Cardiovascular Disease Events (PREVENT) equations to guide the initiation of antihypertensive medications for primary prevention of cardiovascular disease (CVD). This study aimed to compare population sizes recommended for antihypertensive treatment according to primary prevention criteria from the 2017 versus 2025 guidelines.

### Methods 
Using NHANES data (2011–2020), we identified participants aged 30–79 years who had systolic blood pressure (SBP) of 130–140 mmHg or diastolic blood pressure (DBP) of 80–90 mmHg, and who had no prior diagnosis of hypertension, CVD, diabetes, or chronic kidney disease, and were not pregnant nor taking antihypertensive medication. We estimated the 10-year risk for each participant using the Pooled Cohort Equations (PCE; stroke and myocardial infarction for those aged 40–75 years) and the PREVENT base model for total CVD (stroke, MI, and heart failure for those aged 30–79 years). High risk was defined as a 10-year ASCVD risk ≥10% with PCE and total CVD risk ≥7.5% with PREVENT. Analyses incorporated NHANES sampling weights to generate nationally representative estimates.

### Results
The initial population for both analyses included 29.6 million U.S. adults with SBP 130–140 mmHg or DBP 80–90 mmHg, not on antihypertensive medication, not pregnant and without CVD, diabetes, or chronic kidney disease. Of these, 18.4 million were aged 40–75 years and thus eligible for risk assessment with PCE; among these, 3.3 million had a 10-year ASCVD risk ≥10%. In contrast, 24.0 million were aged 30–79 years and thus eligible for total CVD risk assessment using the PREVENT model; of these, 3.5 million had a 10-year PREVENT base total CVD risk ≥7.5%.

### Conclusion
Under the updated thresholds, PREVENT identified nearly 200,000 more high-risk U.S. adults with untreated stage 1 hypertension than PCE, despite differences in model inputs and age range.



Comparison of the 2017 Pooled Cohort Equations and 2025 PREVENT Risk Prediction Models in U.S. Adults with Untreated Stage 1 Hypertension and Without CVD, Diabetes, or CKD: NHANES 2011 to March 2020

### Background
Progesterone receptor membrane component 2 (PGRMC2) is critical for ciliogenesis and the biogenesis of ciliary extracellular vesicles (ciEVs), which play an essential role in maintaining cardiac function. Extracellular vesicles (EVs) are currently being evaluated in eight clinical trials as therapeutic agents for cardiac remodeling. While these studies highlight the therapeutic promise of EVs, the role of ciEVs and their regulation via PGRMC2 remains largely unexplored. This study investigates the impact of cardiac-specific PGRMC2 knockout on heart structure and function and proposes ciEV supplementation as a novel strategy to target cardiac dysfunction, expanding the therapeutic potential of EV-based cardiac therapies.

### Methods 
A cardiac-specific PGRMC2 knockout (KO) mouse model was generated using the Cre-LoxP system to investigate the role of PGRMC2 in cardiac structure and function. Mice were randomized into four groups (n=24 total; equal sex ratio): (1) wild-type with saline, (2) wild-type with EV treatment, (3) KO with saline, and (4) KO with EV treatment. Starting at two weeks of age and continuing through 17 weeks, mice received weekly intravenous injections of either 200 µL saline or 1.6 million EVs.
Assessments were conducted weekly and included treadmill exercise testing and blood pressure measurements. At 18 weeks, cardiac tissue was collected for histological evaluation using Masson’s trichrome staining to assess fibrosis, and left ventricular wall thickness. Statistical significance was defined as P < 0.05.

### Results
Survival analysis demonstrated a high mortality rate (50%) in knockout mice and treadmill exercise-stress tests revealed a marked reduction in exercise capacity (p<0.00001) in KO mice. Blood pressure analysis showed a greater decline in diastolic (p<0.0001 before exercise; p<0.00001 after exercise) and systolic (p<0.0001 after exercise) blood-pressure in male KO mice compared to females.
Pressure-volume (PV) loop analysis assessed cardiac performance under baseline (saline), increased workload (epinephrine), and reduced workload (diltiazem) conditions. KO mice exhibited reduced stroke-volume, indicating impaired cardiac output and efficiency.
Histological analysis revealed left ventricular hypertrophy (p<0.0001 female; p<0.001 male) and a significant increase in cardiac fibrosis (p<0.001 female; p < 0.00001 male) in knockout mice. EV treatment significantly improved survival rate (75%), blood-pressure regulation, ventricular thickness, and cardiac fibrosis.

### Conclusion
Loss of PGRMC2 disrupts ciEVs release, leading to structural and functional cardiac deterioration. Extracellular vesicle injections rescue these deficits, highlighting the essential role of ciEVs and PGRMC2 in maintaining cardiac homeostasis, further reinforcing their therapeutic potential.

PGRMC2 Knockout Disrupts Ciliary Extracellular Vesicle Biogenesis and Cardiac Functions

### Background
One in five adults will develop skin cancer in their lives. Medicare insured adults ages ≥65 account for nearly half of all skin cancers, with a disproportionately higher incidence in rural communities where patients face significant barriers in access to dermatologists. This study aims to fill a gap in the research for this vulnerable patient population by quantifying the relationships between skin cancer prevalence, beneficiary‑service volume of preventative and treatment services, and health care costs for Medicare beneficiaries across the rural-urban continuum from 2019 to 2023.

### Methods
Retrospective analysis of the publicly available 2019-2023 Medicare claims data files and the 2019-2023 National Health Interview Survey. Healthcare Common Procedure Coding System (HCPCS) codes were categorized as skin cancer prevention and treatment services based on the American Medical Association’s CPT Consumer Friendly Descriptors.

Zip codes were cross-walked to four levels of the 2013 National Center for Health Statistics (NCHS) Urban-Rural Classification Scheme: large central metro (LCM), large fringe metro (LFM), medium and small metro (MSM), and non-metropolitan areas (non-metro).

### Results
4,314 Medicare beneficiaries reported a skin cancer diagnosis in their lifetime (8.8%), with a greater incidence of each type of skin cancer in rural areas than urban areas (p<0.03). 

Non‑metropolitan areas had a greater share of dermatology beneficiary‑service volume for preventive services (LCM 23.4%; LFM 23.6%; MSM 25.7%; non‑metro 26.7%; P < 0.001) and services were 2-fold more likely to be any skin cancer treatment and 3.5-fold more likely to be radiation‑related treatment (OR 2.00 [1.99–2.01]; 3.52 [3.48–3.57]) than in LCM areas. 

After standardization for geographic differences in service payment rates, the average Medicare payment for surgical skin cancer treatment services was greater than for preventative care services, and both were greater in rural areas than urban areas (Overall, $118.54 vs. $48.22; LCM, $118.32 vs. $46.85; LFM, $121.7 vs. $48.41; MSM, $117.34 vs. $48.99; non-metro, $116.36 vs. $53.05; p<0.001 for each). Interestingly, no rural-urban difference was observed in radiation-related or chemotherapy treatment services (p=0.230, p=0.260).

### Conclusion
Rural-urban differences in dermatologists’ skin cancer prevention and treatment beneficiary‑service volume reflect a greater need for preventative services in rural areas of the country and provide more recent, service-level evidence for this vulnerable population.

These findings suggest preventative services for skin cancer may significantly lower health care costs in both the short-term and long-term for Medicare patients and CMS, and these calculations may be useful when evaluating new policy for expanding skin cancer prevention efforts nationwide.

Rural-Urban Skin Cancer Prevention and Treatment Differences for Medicare Beneficiaries, 2019–2023

### Background
Bankart lesions (anteroinferior glenoid labrum tears) can cause significant shoulder pain and dysfunction but are often difficult to detect on standard non-contrast MRI. To improve diagnostic sensitivity, MRI arthrography (MRA) is frequently used, although it is more invasive, costly, and uncomfortable for patients. Deep learning (DL), a subset of machine learning, identifies complex patterns in large datasets without the need for manual input. DL has improved diagnostic accuracy across several areas of medical imaging and is particularly well suited for detecting subtle abnormalities. This study developed a DL model to detect Bankart lesions on both MRI and MRA and evaluated its clinical utility through a clinician user study.

### Methods
A dataset of 586 shoulder MRIs/MRAs from 546 patients who underwent shoulder arthroscopy within one year of imaging was retrospectively analyzed. Arthroscopic findings served as the reference standard. A custom neural network ensemble was trained on 410 scans (224 MRI, 186 MRA), tuned on 59 (40 MRI, 19 MRA), and tested on 117 (71 MRI, 46 MRA). Gradient-weighted class activation maps (Grad-CAM) were used to assess the anatomical focus of model predictions. To evaluate clinical impact, a user study was conducted with two shoulder/elbow fellowship-trained orthopaedic surgeons and two orthopaedic residents. Each clinician reviewed all 117 test MRIs/MRAs in two phases: once without model predictions (unaided), and again after a 60-day washout period with DL predictions shown (aided).

### Results
For standard MRI, the model achieved 90.1% accuracy, 83.3% sensitivity, and 90.8% specificity—substantially outperforming radiology reports from the same scans (80.3% accuracy, 16.7% sensitivity, 86.2% specificity). On MRA, the model achieved 89.1% accuracy, 94.1% sensitivity, and 86.2% specificity, compared to radiology reports at 84.8% accuracy, 82.4% sensitivity, and 86.2% specificity. Grad-CAM showed consistent model attention on the anterior labrum. In the user study, mean clinician sensitivity improved from 38.0% to 78.3% with model assistance. Specificity slightly decreased (86.7% to 85.4%), while accuracy increased (77.1% to 84.0%). Clinician confidence increased by 0.65 points on a 10-point scale (p < 0.001).

### Conclusion
The model performed comparably on standard MRI to radiologists on MRA, with Grad-CAM confirming that predictions were based on appropriate anatomic regions. The user study demonstrated that clinicians achieved greater sensitivity and confidence when aided by the model. These results highlight the potential for DL tools to close the gap between non-contrast and contrast-enhanced imaging—avoiding the added cost, discomfort, and invasiveness of MRA.

AI-Assisted MRI Interpretation Improves Surgeon Performance in Diagnosing Bankart Lesions

### Background:
Central blood pressure (BP) more accurately reflects the loading conditions of the heart compared to brachial BP. Brachial BP is routinely used as a surrogate for central BP due to its ease of access, lower cost, and the efficiency of measurement. However, using brachial BP as a surrogate for central BP when calculating carotid arterial stiffness (CAS) has not been studied in older adults.

### Methods:
Veterans (n=180) age 65+ were recruited from Madison VA Hospital. Resting supine brachial BP and central BP (estimated from radial artery waveforms, Atcor Medical) were obtained. CAS (Peterson’s elastic modulus [PEM], Young’s elastic modulus [YEM]) and distensibility coefficient (DC) were calculated using brachial and central BP. Differences in CAS were compared using paired Wilcoxon tests. Linear regression models evaluated associations with cardiovascular risk factors (age, sex, hypertension status, diabetes, sleep apnea, alcohol intake, active smoking, and lifetime smoking status).

### Results:
Participants were 70.4 (7.7) years old and 27.8% were female. Average brachial systolic BP and pulse pressure were significantly higher than central (132.3 [18.6] mmHg vs 123.8 [17.7] mmHg, p<0.001) and pulse pressure (53mmHg [16.4] vs 43.8mmHg [15.3], p<0.001). Compared to brachial BP, using central BP to calculate stiffness measures resulted in significantly lower YEM and PEM and significantly higher DC (PEM: 480.6 [209.5] mmHg vs 378.3 [178.4] mmHg; YEM: 2220.2 [926.6] mmHg vs 1746.9 [785.4] mmHg; DC: 2.4 [1.0] x10-3 mmHg-1 vs 3.1 [1.1]
x10-3 mmHg-1; all p<0.001). Absence of hypertension was associated with smaller differences in PEM and DC (PEM: 𝛽=-29.1, SE=12.1, p=0.02; DC: 𝛽=-123.8, SE=55.3, p=0.027), while older age was associated with greater differences in YEM when calculated using brachial vs central BP (𝛽=1.9x10-5, SE=0.69x10-5, p=0.006).

### Conclusion:
Brachial and central BP significantly differ in older adults and result in significant differences in calculated CAS and distensibility. Brachial BP tends to significantly overestimate CAS, especially in those with hypertension. Our study highlights the importance of considering central BP when evaluating CAS to predict cardiovascular health and better understand vascular aging, especially in older adults with hypertension. Relying solely on peripheral BP may lead to an overestimation of vascular risk in these populations. Future studies should prioritize using central
BP measurements or CAS indices that are independent of BP to more accurately assess arterial health and improve CVD risk stratification. This can help better guide antihypertensive therapy, potentially improving outcomes and reducing the risk of over- or under-treatment.

Central and Brachial Pressures: Effects on Arterial Stiffness in Older Adults

### Background
As large language models (LLMs) gain multimodal capabilities to interpret both images and text, their potential to assist in clinical decision-making continues to grow. While prior research has shown LLMs can match and even outperform individual physicians on standardized exams, little is known about their performance relative to collective human reasoning where varied perspectives contribute to more accurate conclusions. This study evaluates the performance of state-of-the-art multimodal LLMs in solving complex medical challenges compared to both individual and aggregate responses from a large pool of medically literate readers.

### Methods
We compiled 100 ophthalmology-related cases from the New England Journal of Medicine (NEJM) Image Challenge archive (2006–2024), each containing a clinical vignette, image, and multiple-choice question. Six multimodal LLMs (GPT-4 Turbo, GPT-4o, GPT-4o mini, Gemini 1.5 Flash, Gemini 1.5 Pro, and Claude 3.5 Haiku) were prompted to answer each case and self-report confidence (1–4 scale). Human performance was assessed using (1) the average accuracy of NEJM respondents and (2) the accuracy of the most commonly selected (modal) answer per case, representing collective judgment. LLM accuracy and confidence were compared using Kruskal-Wallis and post hoc Dunn’s tests (p < 0.05 considered significant).

### Results
All LLMs except Claude 3.5 Haiku outperformed the average individual respondent (mean human accuracy: 50.4%, 95% CI: 40.8–60.0%), with GPT-4o achieving the highest model accuracy at 72.0% (95% CI: 66.7–76.8%). However, the collective human response—the most frequently selected answer—was correct in 93.0% of cases (95% CI: 86.3–96.6%), significantly outperforming every LLM. The most popular response across LLMs (i.e., the answer most frequently selected by the models) was correct in 65.3% of cases (95% CI: 56.0–74.7), significantly lower than the collective human accuracy. Correct answers were associated with higher confidence scores for all LLMs except Gemini Flash, suggesting self-rated confidence may be a useful proxy for model reliability.

### Conclusion
Multimodal LLMs can outperform individual human respondents on image-based diagnostic tasks but fall short of the accuracy achieved by collective human reasoning. This underscores a key strength of human cognition: the diversity of thought across individuals produces more accurate outcomes than any single decision-maker, human or machine. In contrast, LLMs often rely on a uniform reasoning process that can propagate the same errors consistently. While LLMs hold promise as diagnostic aids, further refinement may be required for them to match the collective intelligence of human experts in interpreting clinical vignettes.

Accuracy of Multimodal Large Language Models vs. Human Consensus in Interpreting Clinical Vignettes

### Background
Infertility is a rising health concern that affects approximately 13% of reproductive age women, but advancements in treatment are limited by our understanding of endometrial function. Decidualization is the process of endometrial stromal cell (eSC) differentiation from a fibroblastic state to a secretory state that occurs during the mid-secretory phase of the menstrual cycle and is associated with a dramatic influx of uterine natural killer (uNK) cells to the endometrium. Defects in decidualization and uNK cell function are associated with infertility, implantation failure, and pregnancy loss, but mechanisms of eSC-uNK cell interaction are not well defined. In this study, we developed and validated uNK cell isolation and expansion techniques using menstrual effluent (ME) as a source of endometrial tissue, then used these ME-derived uNK cells to investigate interactions between uNK cells and decidualizing eSCs to further understand cellular processes occurring prior to embryo implantation.

### Methods
ME samples were collected from participants of the Research OutSmarts Endometriosis (ROSE) study at the Feinstein Institutes for Medical Research. uNK cells were purified by sequential magnetic isolations and expanded in vitro using IL-15 and IL-2. These ME-derived uNK cells were transcriptionally verified by scRNAseq and compared to reference datasets. For co-culture experiments, eSC decidualization was induced by cAMP + MPA. Interactions between uNK cells and decidualizing eSCs were characterized by live-imaging microscopy, eSC staining, and scRNAseq. Functional mechanisms of uNK-eSC interaction were described by protein and gene expression using ELISA and qPCR analyses.

### Results
This novel uNK cell isolation method achieved a purity of up to 92% CD56bright CD16- uNK cells. Cultured ME-derived uNK cells exhibited up to 30-fold expansion, maintained viability for approximately 30 days, and retained phenotypes of reference decidua-derived uNK cells as assessed by flow cytometry and single cell RNA sequencing. In an ME-derived model of uNK-eSC interaction, uNK cells mediated contact-dependent eSC remodeling during decidualization as determined by live-imaging microscopy and cell density analysis. uNK cells concurrently stimulated a pro-inflammatory eSC response in part via IFN release as determined by transcriptomic and proteomic analysis.

### Conclusion
These results indicate that ME is a viable, non-invasive source of uNK cells that can be used for downstream analyses and suggest that uNK cells mediate both eSC remodeling and inflammation during decidualization. Interactions between uNK cells and eSCs are critical for endometrial receptivity, and this novel ME-derived system greatly expands the available tools to investigate mechanisms contributing to endometrial homeostasis and female fertility.

Waste to Wonder: Understanding endometrial interactions using a menstrual-effluent derived model

### Background
Veterans of the armed forces are the embodiment of our nation's security and freedom. The health challenges they face are often exacerbated by service-related issues including psychological trauma, excess repetitive noise, radiation exposure, and subjection to dangerous chemicals from Agent Orange, burn pit smoke, water contamination, asbestos, and more. [ref PMID: 34070145] Veterans experience a higher prevalence of complex health conditions, including increased risk of heart and lung diseases, diabetes mellitus, and various cancers. [ref PMID: 39696828] Providing high-quality equitable care to veterans requires targeted healthcare strategies that address their specific needs.

### Methods
To enhance the quality of care provided to our veteran population at Geisinger, we integrated inquiries about military service and health requirements into an 8-question “About Me” questionnaire. This initiative aimed to establish a clearer connection between veterans and the medical services they require. Utilizing the SlicerDicer® tool in the Epic electronic health record system, we analyzed our veteran population to identify prevalent health risks with the goal of using these data to develop targeted care programs.

### Results
Veterans (n=35,648) had higher overall risk scores for various health conditions compared to the non-veteran population (n=507,773). Specifically, the risk ratios for chronic obstructive pulmonary disease (COPD) were 2.09, for hypertension 1.60, for heart failure 2.47, for diabetes 1.74, and for other cancers 2.19. Among Vietnam veterans (n=5,441), the risk ratios for certain conditions indicated increased vulnerabilities: hypertension at 1.79, type 2 diabetes at 1.99, prostate cancer at 2.89, skin cancer at 2.44, and atherosclerosis at 2.18. Gulf War veterans (1990-present, n=7,586) showed significantly higher relative risks for post-traumatic stress disorder (PTSD) at 7.00, COPD at 3.55, hypertension at 2.84, asthma at 1.58, and lung cancer at 2.14.

### Conclusion
Our findings highlight the critical need for tailored healthcare initiatives for Vietnam and Gulf War veterans, who are the most at-risk populations among veterans. These findings support the development of specialized screening, prevention, and treatment protocols that recognize both the physical and psychological effects of military service as seen in previous studies. This information will be used to create a value-based care guide aimed at optimizing care for our veteran population, as many utilize community healthcare facilities outside of the VA, and this fills an urgent gap in educating healthcare teams about veteran health needs.

Premium content

Downloads

Next from AMA Research Challenge 2025

A proteomic signature of vascular disease is associated with the aging brain's vulnerability to degeneration preceding accumulation of amyloid pathology

Similar lecture

"Rough Day, Need a Hug": Learning Challenges and Experiences of the Alzheimer's Disease and Related Dementia Caregivers on Reddit