Canada

On average, patients with adult-onset Krabbe disease have
a life expectancy of five to seven years after the initial diagnosis.
This rare type of leukodystrophy causes rapid degeneration
of the nervous system, leading to the progressive decline of
neuromotor functions and the appearance of white matter
hyperintensities (WMH) in the brain. Patients with Krabbe
disease experience an increasing impact on their quality of
life as the disease progresses. Therefore, it is vital for a patient
with Krabbe disease to receive optimal care to increase their
life expectancy and receive symptom-specific treatments
on time. Our objective is to create an AI system that helps
physicians and researchers plan the most optimal course of
treatment for patients based on their unique symptoms and
disease progression. Using Jupyter, we created AI models
using Neural networks and Decision Tree Classifiers that use
existing datasets on Krabbe disease to analyze correlations
between symptoms, WMH, and genetic mutations. Our
models can classify the WMH pattern as classical or abnormal
based on symptom frequency with 100% accuracy, further
classify abnormal patterns into extensive or patchy based
on symptoms with 100% accuracy, and predict WMH pattern
based on mutations present with 50% accuracy. This AI
system provides the framework for patient care optimization
by narrowing down optimal treatment options based on the
known WMH pattern and its correlated symptoms.

Leveraging Open Data Analytics And Machine Learning To Improve Diagnosis Of Rare Diseases, Patients’ Care And Support

Improving Rare Disease Diagnosis: Using AI to predict mutations and white matter hyperintensities in adult-onset Krabbe Disease

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

AMA Research Challenge 2025 

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).

technical paper

The Inter-University Big Data Challenge (IUBDC) is an inquiry-driven experiential learning program that invites graduate and undergraduate students from across the country to strengthen their problem-solving and critical thinking skills while gaining familiarity with the fundamentals of data science. The IUBDC fosters scientific inquiry, pioneering thinking, and a solution-focused mindset to drive innovative ideas in the field of the challenges’ topics.

National Inter-University Big Data And AI Challenge 2023 - 

Please find student Open Data and computational science and machine learning finalist projects that are sustainable by design to enhance our understanding of rare diseases, help people affected by the disease to adapt to their lifestyle and reduce their risk of suffering from the disease.

Multiple Sclerosis (MS) is an autoimmune
disease for which there is no existing cure.
Diagnosis of MS involves a clinical assessment
of patients as well as an analysis of the dissemination of lesions through space and time
by imaging exams. However, difficulties in
the proper diagnosis and prediction of disease
progression of MS still lie in the greater discrepancy found in lesions among patients. As
such, in this study, a Machine Learning (ML)
prediction model has been created to assess the
feasibility of ML in predicting lesions in patients
with clinically definite MS, starting a discussion
on whether lesions can be detected/predicted
based on external features. Results have shown
that a model can be created with limited data
with an accuracy of 0.74 with features such
as Final Expanded Disability Scale score and
age of patients playing a significant role in
lesion detection. Limitations of this study
include small sample size, limited number of
features, and restricted information on the
dataset utilized for analysis. Future studies
are recommended for retraining the model in
larger datasets on MS patients as well as adding
additional features for training.

Infratentorial Lesions in patients with Clinically Definite Multiple Sclerosis: A prediction model

Pancreatic Ductal Adenocarcinoma (PDAC) is a highly lethal malignancy in
which early and accurate staging is paramount for effective therapeutic decision-making and prognostication. Traditionally, the prediction of
PDAC staging has depended heavily on invasive methods such as CT scans and tissue biopsies. These methods can be costly, time-consuming, and distressing for patients [1]. Recently, focus has shifted towards the development of non-invasive detection methods utilizing Machine 
Learning. Research conducted by the Barts Cancer Institute[2], which investigated the early detection of PDAC using bio-markers such as
CA19-9, LYVE1, REG1B, REG1A, TFF1, and creatinine, has shown promising results.
Our study is a subsequent investigation that relies on the same open-access data-set comprising 590 participants. It aims to first ascertain 
the previous work of researchers related to early detection of PDAC. Second, to investigate the possibility of utilizing the same bio-markers to
predict the stage of PDAC, this differs our work from previous work.
We have tried out several machine learning algorithms. Our results for the first problem, early prediction of PDAC are very promising - we achieved an accuracy of 87%, 84% and 78% for precision and recall respectively utilizing
XGBoost. For the second problem, predicting the stage of PDAC, MLP neural networks yielded an accuracy of 70% with an AUC of 79%. These results are promising and better than the base models, which opens a new
area for future works. Based on our work, we propose the establishment of a certified pancreatic bio-marker platform.

Predicting the stages of PDAC using non-invasive method

Breast cancer is one of the leading cancers worldwide and by 2040, it is expected to
cause more than 1 million deaths. Hence it is important to study the cytology of this
cancer to track disease progression. In this research, we have utilized a dataset from
the Wisconsin breast cancer dataset which contains 699 samples of benign and
malignment tumors. The samples contain clump size, cell size, and shape, single
epithelial cell size, chromatin, nuclei, nucleoli, mitosis, and adhesion. We have
constructed a data science pipeline through data collection, exploratory data analysis,
and building the machine learning model. After collecting the data, we explored the data
to see the relationship between the independent variables by constructing different plots
such as the pair plot, violin plot, and heatmap. Then we built five different supervised
machine learning models such as the decision tree, naive Bayes algorithm, deep
learning, support vector machine, and logistic regression with associated parameters
and hyperparameters. Then the model has been evaluated using different evaluation
metrics, for example, the area under the curve, accuracy, sensitivity, and specificity
scores to compare them. Receiver operating characteristics plots also have been
constructed to investigate the prediction ability of those models. According to the
results, deep learning and support vector machines with all features have the highest
AUC score which is 0.99 and 0.998 and this indicates these models can predict breast
cancer progression more accurately.

“Construction of a data science pipeline and comparing different supervised machine learning algorithms to predict breast cancer”

Rare diseases are a prevalent problem within the healthcare sector, largely attributed to the multitude of conditions yet lack of profiling for individual diseases. Consequently, the development of novel therapeutics for rare diseases through traditional pipelines is fraught with risk and cost. Integration of machine learning within bio-informatics has facilitated the growth of in silico drug repurposing, wherein the discovery of novel molecular targets for compounds enable their repositioning for treating new conditions.
The aim of this study is the proposal of a multi-class predictive model capable of accurately determining the mechanisms of action for existing compounds. This, in tandem with limited rare disease profiles, is further applied to generate candidates for drug repurposing.
Four supervised learning algorithms were utilized, trained using a combination of Clue and PubChem datasets. Results indicate the
random forest model boasts the best performance with an accuracy of 0.724, comparable to existing literature. As all the top-performing
models in this study were black-box, interpretability features of LIME and OpenAI API were integrated to ensure transparency in the recommendation of candidates, preparing the model for healthcare deployment. This model
was then applied through a case study of Infantile Dystonia-Parkinson (IDP), where 105,552 compounds were screened for association with the condition. This study accurately proposes pramipexole, bromocriptine, and ropinirole as candidate compounds for further investigation for IDP treatment. Due to the accuracy of both mechanism prediction and candidate generation, this computational model stands as a suitable approach for rare disease drug repurposing.

A Needle in a Haystack: Leveraging Machine Learning for Drug-Mechanism of Action Identification across Existing Therapeutics, with Specific Applications for Drug Repurposing for Rare Diseases

Diagnosing rare diseases often presents significant challenges due to their complex, unique characteristics, leading to delays in appropriate interventions and impacting patient outcomes. Addressing this issue, we introduce the ‘DiagnosisAssistant,’ an innovative web-based application designed to assist medical professionals in efficiently and accurately di-
agnosing rare diseases. Integrating artificial intelligence and machine learning, DiagnosisAssistant facilitates the recognition of Human 
Phenotype Ontology (HPO)[1] concepts in unstructured medical text. This integration made possible through the PhenoTagger[2] tool, enables healthcare professionals to input patient symptoms and signs in a natural, descriptive manner, streamlining the diagnostic process.
Furthermore, the application leverages the comprehensive HPO and Orphanet[3] databases, crucial resources in rare diseases. Our performance testing on ten varied case reports of rare diseases has shown promising results, reinforcing the app’s potential in complex diagnostic processes. Despite its dependency on the completeness and accuracy of the HPO and
Orphanet databases and its current inability to account for patient-specific factors such as age, medical history, and genetic data, the tool represents a significant step forward in rare disease diagnostics. With its source code
openly accessible for continuous improvement, the DiagnosisAssistant holds immense potential for transforming the diagnosis of rare diseases,
thus significantly enhancing patient care and treatment outcomes.

Diagnosis Assistant: A Web-Based Application for Assisting in Diagnosing Rare Diseases

**Background:** Long COVID is the long-term health impact of COVID-19 infection. The
Omicron variant has increased the risk and prevalence of Long COVID, raising public concerns
and awareness.
<br>**Objective:** To investigate the online discussion and sentiment on Long COVID on Twitter over
time in relation to the COVID-19 cases in the US and globally and to identify the main topics
and themes of discussion across different periods of the pandemic.
<br>**Methods:** Twitter data from January 1, 2020, to December 31, 2022, were collected and
analyzed using machine learning algorithms, Pearson correlation analysis, sentiment analysis,
and topic modeling. The data were divided into three periods based on the discovery and impact
of the Omicron variant: pre-Omicron, Omicron, and post-Omicron.
<br>**Results:** The volume and sentiment of Long COVID tweets were positively correlated with
COVID-19 cases, indicating that social media can monitor the spread and severity of Long
COVID. The topics of discussion changed over time, reflecting different aspects and impacts
of Long COVID on public health and well-being. Four main topics were identified: scientific
research, financial and political issues, mental health, and vaccination.
<br>**Conclusions:** The paper provides novel insights into the online discourse and sentiment on
Long COVID on Twitter during the pandemic. The paper suggests implications for public
health communication and intervention strategies during health crises.

Online Discourse and Offline Health Contagion: Longitudinal evidence from Long COVID on Twitter

Medical text in natural language is an important source of biomedical knowledge, complementing clinical numeric data. How to effectively study the patterns between different diseases is important to medical advancements. With significant achievements in language modeling tasks in recent years using deep neural networks approach, a wider array of topic models has become available. To more effectively study the relationships between rare diseases and their symptoms, we apply conventional statistical models such as LDA and a newer embedding model BERTopic for topic modeling of texts containing description of rare disease symptoms. We compare the performance of topic models on the learned topics by Topic Coherence(TC) and Topic Diversity(TD). We also evaludate their effectiveness for keyword-based documents, demonstrating the patterns that exist in rare disease symptom descriptions.

Topic Modeling for Rare Disease Symptoms

Parkinson’s disease (PD) is a chronic neurodegenerative condition characterized by the degeneration of nerve cells in the brain, leading to motor symptoms such as tremors and rigidity. While the exact cause of PD is not fully understood, studies have shown the involvement of specific proteins, including alpha-synuclein.
This paper explores the predictive potential of proteins in PD using data derived from mass spectrometry readings of cerebrospinal fluid (CSF) samples. The goal is to identify proteins that may have relevance in early detection and personalized treatments for PD. The study utilizes machine learning techniques, including regression and classification models, to establish the relationship between normalized protein expression (NPX) and PD progression. The results indicate that there is no significant relationship between NPX and PD progression based on the analyzed approaches. However, limitations in the study, such as the absence of complete clinical information and the need for imputation, must be considered. Further research is required to gain a more comprehensive understanding of the complex molecular mechanisms associated with PD.

Exploring the relationship between Protein Expression in Cerebrospinal fluid and Parkinson’s Disease Progression using Machine Learning

Overwork and understaffing put a great strain on nurses around the globe. With healthcare workers suffering from burnouts, patients have a difficult time connecting with professionals when they need it most. This paper aims to investigate the potential of creating a highly accessible and cost-efficient virtual assistant able to support the health sector by interacting with patients and recording medical information from them. To create this interactive virtual assistant, a speech recognition model, a language model and a text-to-speech software were selected and then combined to make Pythia, a prototype virtual assistant on which tests can be conducted. A text classification model was also developed to sort out medical conversations
that may prove useful to the patient’s record and save them in an online database under the patient’s name. Tests were conducted on Pythia to ensure its understanding of medical conversations and its ability to respond rapidly and accurately to questions. Utilising OpenAI’s Whisper and GPT-3 alongside ElevenLabs’ text-to-speech software, Pythia demonstrates
that it is possible to create an interactive virtual assistant with accessible online resources and the result may be adapted to the health sector
as a means to alleviate workloads. The current prototype functions best in a quiet environment and has difficulty diverging from medical conversations. While the text classification model requires further development, the virtual
assistant itself is functional and able to gather symptoms from patients. Further work on the subject is however necessary to ensure the
virtual assistant’s practicality and expertise on the field.

Downloads

Next from Leveraging Open Data Analytics And Machine Learning To Improve Diagnosis Of Rare Diseases, Patients’ Care And Support

Infratentorial Lesions in patients with Clinically Definite Multiple Sclerosis: A prediction model

Similar lecture

Understanding the Impact of Race on White Matter Hyperintensities in Alzheimer’s Disease: Findings from the ADNI Study

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from Leveraging Open Data Analytics And Machine Learning To Improve Diagnosis Of Rare Diseases, Patients’ Care And Support

Infratentorial Lesions in patients with Clinically Definite Multiple Sclerosis: A prediction model

Similar lecture

Understanding the Impact of Race on White Matter Hyperintensities in Alzheimer’s Disease: Findings from the ADNI Study

Downloads