United States

Hereditary hemochromatosis (HH) is a 
genetic disorder characterized by iron 
overload, which can lead to elevated 
liver enzymes, increased ferritin and 
transferrin saturation. The condition is 
associated with a heightened risk of 
liver and metabolic complications, 
including fatigue, joint pain, abdominal 
pain and endocrinopathies. 
Emerging preclinical data suggest that 
GLP-1 receptor agonists (GLP-1RAs) 
may improve hepatic inflammation and 
iron metabolism in HH.

AMA Research Challenge 2025 

Beyond Phlebotomy: Tirzepatide for Iron Overload in Hereditary Hemochromatosis

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. **

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).

poster

Non-small cell lung cancer (NSCLC), constituting about 85% of lung cancer cases, has a notably high incidence in Kentucky. Recent advancements have improved 5-year survival rates from a dismal 9.7% to approximately 18.4% with the introduction of combination chemotherapy and immune checkpoint inhibitors (ICIs) in 2018. Despite these gains, resistance to this combined therapy remains a significant issue. A proposed mechanism for such resistance involves CD73+ myeloid-derived suppressor cells (MDSCs), which may be induced by prostaglandin E2 (PGE2) accumulation following chemotherapy.
This study investigates the pathway through which platinum-based chemotherapy agents induce PGE2 overproduction, leading to the development of CD73+ MDSCs. We hypothesize that chemotherapy increases fatty acid synthase (FAS) activity, which elevates arachidonic acid (AA) levels. AA is a substrate for PGE2 synthesis, and its increased availability stimulates the NF-κB/CREB pathway, enhancing COX-2 activity and PGE2 production. The accumulated PGE2 is then thought to drive the formation of CD73+ MDSCs, which contribute to immunotherapy resistance by producing adenosine that suppresses T-cell activity.
Our findings, based on mass spectrometry, western blot analysis, and ELISA assays, demonstrate that chemotherapy significantly increases AA levels and activates the enzyme cascade responsible for PGE2 production. Mass spectrometry showed a doubling of AA concentration in chemotherapy-treated LLC cells compared to controls. Western blot analysis revealed enhanced FAS, NF-κB, and CREB activity in chemotherapy-treated KP cells. ELISA results indicated a marked increase in PGE2 levels in the supernatants of cisplatin-treated cells, even with PTGES knockdown, confirming the link between chemotherapy and elevated PGE2 levels.
Additionally, flow cytometry analysis in NSCLC mouse models revealed that chemotherapy treatment increased the percentage of CD73+ MDSCs from 10% to 20%. These results confirm that chemotherapy drives the overproduction of PGE2 and contributes to the expansion of CD73+ MDSCs, which may play a critical role in the resistance to immunotherapy. These insights could guide the development of targeted strategies to overcome resistance and improve therapeutic outcomes in NSCLC.

Chemotherapy induced accumulation of PGE2 & resulting immunotherapy resistance

Background
Chronic Kidney Disease of unknown etiology (CKDu) is a progressive form of kidney damage that disproportionately affects agricultural communities in Sri Lanka, Central America, and parts of India, with emerging hotspots also suspected in the South and Western United States. Despite extensive investigation, the underlying causes of CKDu remain unclear, partly because of a potential long lag between exposure and eventual symptomatic presentation of kidney disease. We applied machine learning (ML) and large language models (LLMs) to explore predictive tools for disease progression in a cohort of persons with CKDu, with the hypothesis that early identification of persons likely to experience kidney function decline will facilitate investigation of proximal (causative) exposures.
Methods
We analyzed baseline data (labs demographics) from 244 participants enrolled in Kidney Progression Project (KiPP), a prospective cohort of farmworkers with kidney disease in Sri Lanka launched in 2018. Among these participants, 22 progressed to the study’s primary outcome (eGFR<15 mL/min/1.73 m²) by the year 2024. For model development, we used data from the first 2.5 years of follow-up.
We developed and optimized several ML classifiers including logistic regression, support vector machines, random forests, and XGBoost. Each model was trained using stratified five-fold cross-validation, and class imbalance was addressed using SMOTE-based oversampling. To identify underlying patterns in the dataset, we applied unsupervised models like K-means clustering(k=7), and the resulting cluster labels were included as categorical features in supervised learning models. Model performance was evaluated using standard classification metrics, and interpretability was assessed using SHAP (SHapley Additive exPlanations). 
Additionally, we used clinical vignettes from KiPP data to evaluate the predictive capabilities of LLMs. We tested five LLMs through SecureGPT on their ability to predict progression.
Results
The best performing ML model was an XGBoost classifier (ROC-AUC= 0.84; F1 score= 0.48). SHAP analysis identified cluster membership and serum uric acid as key predictors. Among the LLMs, GPT 4.5 performed best (Accuracy= 87%; F1 score of 0.42). Both models outperformed existing methods (eGFR slope, Accuracy=67%, F1 score=0.23). 
Conclusion
Our findings demonstrate that both traditional ML models and LLMs can predict CKDu progression from limited clinical data. Custom-tailored ML models slightly outperformed generative AI models such as GPT-4.5. With continued refinement, these tools may not only support early identification of high risk individuals but also aid in identifying causes of CKDu. Future directions include external validation using ongoing international cohorts and incorporation of longitudinal laboratory data to generate clinically-deployable AI tools.


From Algorithms to Agriculture: Applying AI Methods to Predict Kidney Disease in Farmworkers

Background
Epileptic seizures disrupt normal brain activity, with pathological activity detectable on electroencephalograms. While seizure classification guides diagnosis and treatment, current systems often fail to capture individual variability in drug response, even among patients with similar clinical phenotypes. This dissertation applies dynamotype classification—a framework grounded in bifurcation theory and dynamical systems—to quantify seizure onset and offset patterns and their modulation by antiseizure medications (ASMs).
Methods
Data from over 7500 seizures was analyzed from 100+ mice from two models: the intra-amygdala kainate (IAK) model of temporal lobe epilepsy and the genetic Scn1a+/- model of Dravet syndrome. In the IAK model, spontaneous seizures were tracked over a 90-day period. Dynamotypes were visually scored in a randomized, blinded manner by two human raters and a custom DynamoSort machine learning tool we developed. Seizure onsets were categorized into three dynamotypes: (1) SupH, characterized by increasing spike amplitude (supercritical Hopf); (2) SNIC, by increasing frequency (saddle-node on invariant circle); and (3) SubH, by arbitrary or disorganized features. Seizure offsets were similarly categorized: SupH (decreasing amplitude), SNIC (decreasing frequency), and FLC (fold limit cycle, showing arbitrary declines).
Results
Longitudinal analysis of the IAK model revealed a temporal shift in dynamotype prevalence, with early seizures dominated by arbitrary transitions (SubH, FLC) and later seizures showing more structured bifurcation types (SupH, SNIC). Dynamotypes also predicted seizure duration and severity, with SNIC onset seizures being more severe and FLC offset seizures being longer in duration. Drug trials demonstrated that ASMs selectively modulated bifurcation types in a mechanism-dependent manner. GABAergic drugs (diazepam, phenobarbital) predominantly suppressed seizure onsets, indicating preferential disruption of initiation dynamics. In contrast, sodium channel blockers (phenytoin, valproate) exerted stronger effects on seizure terminations: phenytoin reduced SupH offsets, while valproate increased SNIC offsets in treatment responders, indicating SNIC offset seizures may be pharmacoresistant to VPA. Baseline dynamotype distributions also differed across models. Dravet mice displayed elevated SubH onset proportions relative to IAK, but GABAergic drugs caused similar dynamotype changes, showing conserved responses to ASMs across models.
Conclusion
Overall, this work demonstrates that bifurcation-informed seizure classification captures dynamic features of epileptogenesis and drug response, revealing mechanistically distinct signatures across ASM classes and models. These findings support the use of dynamotypes as a model-driven framework for evaluating treatment efficacy and tailoring therapies in epilepsy.

Seizure Dynamotype Patterns Predict Drug Response in Animal Models of Epilepsy

Oral contraceptive pills (OCPs) are one of the most widely used reversible contraceptive methods in the United States. Since their introduction in the 1960s, OCPs have been prescribed globally for birth control as well as management of hormonally influenced conditions including menstrual irregularities, menorrhagia, dysmenorrhea, acne, premenstrual syndrome, and hormonal migraines. Despite their therapeutic benefits, the effectiveness of OCPs varies between individuals, and the physiological mechanisms underlying these effects remain incompletely understood. While OCPs are known to influence menstrual bleeding patterns, little is known about their impact on the vasculature of the female reproductive system, specifically the uterine and ovarian arteries.
In comparison, intrauterine devices (IUDs) have been studied extensively for their vascular effects. Doppler ultrasound studies, which measure pulsatility index (PI) and resistance index (RI) as indicators of arterial blood flow, generally show no significant vascular changes pre- and post-IUD insertion. However, notable exceptions have been reported between copper and hormonal IUDs. Women experiencing abnormal bleeding with copper IUDs exhibited significant changes in PI and RI; however, another study demonstrated an increased RI in the uterine artery following levonorgestrel IUD (LNG-IUD) placement, potentially explaining its association with reduced menstrual bleeding. Given that copper IUDs tend to increase menstrual bleeding, while LNG-IUDs and OCPs typically reduce it, vascular changes may be an important underlying mechanism. This study aimed to investigate the effects of OCPs on uterine and ovarian artery blood flow using Doppler ultrasound in healthy, nulliparous women and to compare self-reported menstrual symptoms, including pain, cycle regularity, PMS manifestations, and skin changes using a standardized Menstrual Symptom Survey between OCP users and non-users.

The Vascular and Menstrual Symptom Impact of Oral Contraceptives: A Comparative Analysis of Uterine and Ovarian Artery Indices via Doppler Ultrasound

Background: Total hip arthroplasty (THA) alleviates pain and improves mobility in individuals with end-stage hip disease. However, THA patients with neuropathic bladder may face increased postoperative risks. This study aimed to assess how neuropathic bladder conditions influence outcomes after primary THA and propose recommendations to avoid associated complications.

Methods: A retrospective cohort analysis was conducted using data from TriNetX Research Network, including 193,988 patients aged 18 years and older who underwent primary THA between January 1, 2003, and January 1, 2020. Patients were categorized into two cohorts: those with (NB) and without (no-NB) neuropathic bladder conditions prior to THA. Propensity score matching yielded 1,482 patients in each cohort. Rates of postoperative outcomes at 30 days, 90 days, 1 year, and 5 years were compared.

Results: Within 30 days of surgery, NB patients demonstrated significantly elevated rates of urinary tract infection, pneumonia, periprosthetic joint infection (PJI), deep vein thrombosis, and transfusion compared to no-NB patients (P < 0.01). Similar trends continued at 90 days post-THA, with the NB cohort also showing a higher risk of pulmonary embolism and acute renal failure (P < 0.01). Furthermore, the NB cohort faced increased risks of PJI over a 1-year period (risk ratio (RR): 1.8; 95% confidence intervals (CI): 1.3, 2.4) and revision arthroplasty over a 5-year period (RR: 2.6; CI: 1.8, 3.8) compared to the no-NB cohort.

Conclusion: Patients with neuropathic bladder undergoing THA faced significantly elevated risks of postoperative complications, highlighting the need for tailored postoperative management strategies of neuropathic bladder to mitigate these risks and improve THA outcomes.

Assessing the Impact of Neuropathic Bladder on Total Hip Arthroplasty Outcomes:
A Matched Cohort Analysis

Background
COVID-19 has been associated with a broad range of neuropsychiatric sequelae, including new-onset psychosis, mood episodes, depression, anxiety, and cognitive impairment. These symptoms can develop during active infection or emerge in the weeks that follow. The underlying causes are likely multifactorial involving neuroinflammation, immune response, direct CNS effects, and psychosocial stressors. Patients with chronic psychiatric illnesses may be especially susceptible to decompensation but reports of COVID-19 related psychiatric decline in inpatient psychiatric settings remain limited.
Case Presentation
A 59 year old female with schizoaffective disorder, bipolar type, and polysubstance use was admitted for psychosis and catatonia. She initially was stabilized on sublingual buprenorphine, Invega Sustenna, and Ativan. By hospital day 27, she was discharge ready and denied hallucinations, suicidality or depressive symptoms. Shortly after stabilization, she endorsed symptoms of diarrhea, no URI symptoms, and subsequently tested positive for COVID-19. Following the positive COVID-19 PCR test, she was placed in isolation and was monitored for additional symptoms. Serial antigen tests 5 and 7 days after were negative. Within days of the initial positive test, she exhibited acute decompensation characterized by increased paranoia, auditory and visual hallucinations, disorganized behavior, and mania. Despite being on therapeutic doses of antipsychotic medications, she became increasingly labile, exhibited disorganized behavior, and made illogical statements. Management was complicated by suspected akathisia, bradycardia limiting propranolol use, and side effects from Lithium which was discontinued and replaced by Valproic acid for manic symptoms. Trial of Olanzapine was begun but was insufficient. On day 55, Clozapine was initiated due to persistent psychosis and mania. Despite concerns about subjective akathisia, she tolerated the initial titration, and mirtazapine was initiated for adjunctive management of akathisia and mood symptoms. The patient remains hospitalized under close psychiatric monitoring, with ongoing management focused on titration of clozapine and adjunctive pharmacologic support.
Discussion
This case highlights the risk of psychiatric destabilization in patients with severe mental illness following COVID-19 infection, even in the absence of severe respiratory symptoms. Management was further complicated by medication side effects and the need for antipsychotic regimen adjustments during an acute medical illness. Furthermore, careful medical management and early recognition of psychiatric relapse are critical in patients susceptible to COVID-19 in the inpatient psychiatric setting.


Psychiatric Decompensation following COVID-19 Infection in an Inpatient Psychiatric Patient: A Clinical Vignette

Background:
Effective physician-patient communication is essential for building trust, supporting shared decision-making, and ensuring treatment adherence. While language barriers are a recognized challenge, cultural differences, such as beliefs about illness, family roles, and healing practices, also significantly impact patient care. Studies show that cultural awareness may improve patient satisfaction even more than racial concordance by fostering individualized, patient-centered care.
Despite its importance, cultural humility training in medical education is often limited to preclinical years, with minimal reinforcement during residency when learners are most involved in direct patient care. This gap is especially concerning given the growing diversity of the U.S. population. Latinos now comprise nearly 20% of the population and represent the largest ethnic minority group in the country. Expanding resident education to include culturally relevant knowledge and skills is critical to improving care for Hispanic/Latine patients. Our project addresses this need by implementing a focused educational curriculum and evaluating its impact on residents’ comfort, communication, and clinical practice.
Methods:
We designed a 60–90 minute, specialty-neutral, case-based curriculum exploring key cultural values relevant to Hispanic/Latine patients, including familismo, personalismo, respeto, simpatía, and espiritualidad. Sessions were interactive and delivered to residents in Internal Medicine, Pediatrics, OB/GYN, Family Medicine, and Surgery. To assess the curriculum’s impact, we administered pre- and post-session surveys using a shortened, customized version of two validated tools: the Cross-Cultural Care Survey and the Cultural Competence Health Practitioner Assessment.
Results:
Prior to the sessions, 47% of residents reported feeling only slightly comfortable identifying cultural customs that could impact care, and 88% expressed a desire for more training. After the sessions, 55% felt somewhat comfortable discussing how culture influences care. Understanding of Latine health beliefs and use of traditional healers improved significantly, with knowledge of traditional healer use increasing from 19% to 47%. While 45% of residents found the sessions extremely helpful, only 50% felt moderately comfortable incorporating the strategies into clinical practice.
Conclusion:
These findings highlight both a clear need and strong demand among residents for cultural humility education. The curriculum improved knowledge and awareness, but the modest increase in comfort applying the material suggests that single-session interventions may not be sufficient. Longitudinal reinforcement and case-based practice may be necessary to build confidence. Incorporating cultural training alongside language-concordant care holds promise for strengthening physician-patient relationships, improving communication, and enhancing outcomes in cross-cultural encounters.


Cultural Humility in Residency: Latine Patient-Centered Teaching

While professional identity formation is recognized as beneficial for medical students’ well-being and relationships with colleagues and patients, students consistently report challenges with engaging in meaningful respite and reflection. Time spent in nature has been repeatedly shown to support well-being, reduce stress, and enhance reflective capacity, yet remains to be used in a reflective practice for students. In contrast, lack of contact with nature is associated with higher levels of anxiety, depression, decreased creativity, attention difficulties, and decreased overall physical and emotional well-being. This study explores medical students’ reflections on important formative connections to nature and the barriers that limited these connections. 

We collected data from 56 medical students from across the United States who participated in the Backpacking Medicine course offered by the Advanced Wilderness Life Support (AWLS) program at the University of Utah School of Medicine. Participants were divided into four focus groups and asked to write on a notecard their favorite memory in nature and a personal barrier to accessing nature. These reflections were discussed within the groups in relation to students’ current experiences in medical school and their future roles as physicians. The written responses were analyzed using thematic analysis, with codes organized into themes and subthemes. 

Three major themes emerged for students’ favorite memory: Family & Social Bonds, Independent Play & Exploration, and Structured Outdoor Programs & Skill Building. Six major themes emerged for students’ barriers to outdoor access: Health & Physical Limitations, Negative or Traumatic Outdoor Experiences, Lack of Social Support or Modeling, Environmental and Structural Access Barriers, Cultural and Generational Barriers, Time & Life Demands. 

Our findings highlight the role of nature exposure in promoting reflective practices, identity development, and well-being in medical students. Students’ favorite outdoor memories emphasized the value of social connection, skill development, and personal growth. Reflections on barriers to the outdoors prompted students to consider both their own access to nature and that of future patients, inspiring creative thinking around the idea of “prescribing” nature in clinical care. A key limitation of this study is sample bias, as participants signed up for this nature-based course and may have held prior interest in or appreciation for the benefits of outdoor experiences. Future research is needed to determine the long-term effects of implementing nature as a reflective tool in medical education. 





Exploring the Use of Nature as a Reflective Tool in Medical Education

Background: Both language concordance and Medicaid acceptance are essential for equitable surgical care. Spanish-speaking and other multilingual surgeons improve comprehension, satisfaction, and outcomes for patients with limited English proficiency (LEP), while Medicaid participation ensures economic accessibility. Chicago houses more than 1.7 million residents who speak a language other than English—predominantly Spanish—and over 25 % of city adults rely on Medicaid. Whether large academic health systems have enough Medicaid-accepting, Spanish-
speaking, or otherwise multilingual surgeons to meet this dual need remains unclear. The objective was to determine if Medicaid-accepting Spanish-speaking or multilingual surgical specialists are disproportionately underrepresented at the five largest academic institutions in Chicagoland.

Methods: In February–July 2025, publicly available roster data were extracted from surgeons in general surgery, orthopedics, neurosurgery, urology, vascular surgery, otolaryngology, ophthalmology, OB/GYN, and dermatology from Rush, Northwestern, the University of Chicago, Loyola, and UIC. Variables collected included languages spoken, accepted insurance plans, and clinical practice ZIP codes. Surgeons were deemed multilingual if ≥2 languages were listed and Spanish- speaking if Spanish was explicitly reported. Medicaid acceptance was coded when any Illinois Medicaid plan (e.g., CountyCare, Meridian, Molina, MMAI) appeared among accepted insurances. Descriptive statistics compared proportions of (1) multilingual, (2) Spanish-speaking, (3) Medicaid-accepting, and (4) Medicaid accepting + Spanish-speaking surgeons across institutions and specialties.

Results: Among 1087 surgical specialists identified, 204 (18.8%) were multilingual, and 96 (8.8%) spoke Spanish. Medicaid was accepted by 278 providers (25.6%); however, only 38 surgeons (3.5%) met the combined criteria of Medicaid-accepting & Spanish-speaking, and 66 (6.1%) were Medicaid-accepting & multilingual. Representation varied markedly by institution: Rush (5.5 %) and UIC (6.3 %) had the highest share of Medicaid-accepting Spanish-speaking surgeons, whereas Northwestern and the University of Chicago had <3 %. Specialty analysis showed the lowest dual-access representation in neurosurgery (1.8 %) and vascular surgery (2.4 %). Geographic mapping revealed that many Medicaid-accepting multilingual surgeons practiced in affluent downtown ZIP codes, leaving LEP-dense neighborhoods with few or no language-concordant, Medicaid-accessible surgical specialists.

Conclusion: Medicaid-accepting Spanish-speaking and multilingual surgeons are indeed under-represented within Chicago’s largest academic institutions and are unevenly distributed away from high-need LEP communities. Addressing this dual disparity will require institution-specific recruitment incentives, multilingual training pipelines, and strategic placement of surgical services in linguistically diverse, Medicaid-reliant neighborhoods to close persistent gaps in access and outcomes.

Language and Medicaid Access Disparities Among Surgical Specialists at Major Chicagoland Institutions

Lung cancer is a silent killer, often going undetected until it’s too late. Early diagnosis greatly improves outcomes—stage I patients have a 97% two-year survival rate, compared to just 23% at stage IV. However, most diagnoses occur at later stages due to limitations in current screening methods. Current detection methods, namely Gas chromatography–mass spectrometry (GC-MS) and CT scans, are costly or invasive, leading to most diagnoses at advanced stages. These methods require in-lab operation, and are inaccessible to rural or resource-limited settings. While several studies have explored small-scale sensor systems as an alternative, they often suffer from reduced accuracy, high production costs, and long processing times—sometimes exceeding 40 minutes per breath sample. Therefore, there is a need for a fast, affordable, and accurate breath-based lung cancer detection system suitable for those with limited resources. To address this, we developed an affordable breath-based lung cancer screening device featuring a gas chamber and machine learning model. It analyzes VOC composition in human breath, influenced by lipid peroxidation in cancer cells, to compare profiles between healthy and lung cancer samples. The system was designed to detect more than 30 volatile organic compounds (VOCs) using twelve metal oxide semiconductor sensors and one chemi-resistive alkane sensor. In total, 46 clinical breath samples were collected (from 28 healthy controls and 18 lung cancer patients) in 1L Tedlar bags. Electrical resistance data from the sensor array was processed using a multilayer perceptron (MLP) neural network. The model architecture incorporated dropout and batch normalization layers to minimize overfitting. Sensor output patterns were statistically analyzed using the Mann-Whitney U test, which confirmed significant differences between cancer and control groups (p < 0.05). Model performance was evaluated using five-fold cross-validation to ensure robustness and generalizability. The final model achieved an accuracy of 96.57%, sensitivity of 92.88%, and specificity of 98.82%, with an area under the ROC curve (AUC) of 0.9913. Low standard deviation across folds indicated consistent classification performance. The complete system was able to produce diagnostic results in about 10 minutes per breath sample. Our system outperforms existing methods of lung cancer detection through MOS sensors by greater than 5% while costing only around $300. These findings suggest that the breath analyzer system proposed in this study has the potential for preliminary lung cancer screening. 





Premium content

Next from AMA Research Challenge 2025

Chemotherapy induced accumulation of PGE2 & resulting immunotherapy resistance

Similar lecture

Severe Iron Overload Presenting as Pancreatitis in a Patient with Chronically Transfused Sickle Cell Disease