Location - HSRB Rollins Auditorium

1760 Haygood Dr NE, Atlanta, GA 30322

Continental Breakfast & Registration

Debrief & House keeping

Review of the day’s activities and important housekeeping notes.

Finding the Correct Answer (Ground Truths)

Explore the types of labels needed (class, mask, bbox) and their sources (PCR, sequencing, reports, other modalities). Discuss the history of labels, hidden stratifications, label correlation, and alternative approaches for label extraction using segmentation, Imagenome, and generative models.

Break

Models on the CXR Dataset

Understand the end-to-end model pipeline, including data processing, results representation, and potential pitfalls.

Lunch

Inference and Stats on the CXR Dataset

Delve into inference using models, metrics for inference, and statistical analysis.

Break

EMBED: Introduction to Breast Screening and Diagnostic Pathways

An introduction to breast screening and diagnostic pathways.

Introduction to the EMBED Dataset

Overview of the EMBED dataset and its applications.

Location - HSRB Rollins Auditorium

1760 Haygood Dr NE, Atlanta, GA 30322

Continental Breakfast & Registration

Debrief

Review of the day’s activities and important housekeeping notes.

EMBED - Data Engineering

Curate a patient cohort with MagView and metadata, focusing on cancer/no cancer, normal vs. abnormal classifications, and stratification by breast density. Address common pitfalls and create a test set to highlight key issues.

Break

EMBED - Data Engineering: Making Use of ROIs

Learn how to effectively leverage Regions of Interest (ROIs) in data engineering for enhanced analysis and insights.

EMBED - Data Preprocessing

Prepare a dataset for training by performing DICOM preprocessing, stratifying by subgroups, applying VOI LUT, masking breast tissue, standardizing image orientation, and addressing preprocessing challenges.

Lunch

EMBED - Training Whole Image Classifier

Training a whole image classifier or other model for BI-RADS 0 vs. 1/2 categories.

EMBED - Model Evaluation and Inference

Utilize our model for running inference, calculating statistical metrics, and evaluating subgroups. Extract imaging findings and analyze demographics, including race, ethnicity, age, density, and pathology.

Break

EMBED - Model Evaluation and Inference

Utilize our model to run inference, calculate statistical metrics, and evaluate subgroups. Discuss appropriate metrics like F1, AUC, specificity, and sensitivity, along with bootstrapping, class balancing, and potential pitfalls in model performance evaluation.

Location - Goizueta Business School 208

1300 Clifton Road, Atlanta, GA 30322

Continental Breakfast & Registration

Debrief

An overview of bias and unfairness in AI/ML

Prof. Odom

EMBED - Breast MRI Overview

Break

EMBED - Breast MRI Model

Classify MRI BPE, select and extract data from 1000 MRIs. Explore data, use pre-trained 3D models, fine-tune, evaluate, and compare performance of 3D classifiers, stacked image classifiers, middle images, and eigen images.

Lunch

Radiopathomics

Exploring the intersection of radiology and pathology through advanced data analysis techniques.

Humanizing AI: The Patient's Influence in Reducing Bias in Women's Health

In AI-driven healthcare, especially in women's health and maternal care, it's crucial to recognize that clinicians and patients are key stakeholders directly impacted by these technologies. This presentation will highlight the importance of considering the human aspect in developing equitable and effective AI tools.

Break

Bias in AI

Examination of biases present in AI models and their implications in healthcare.

AWS SageMaker Workshop

Learn to utilize AWS Sagemaker for building, training, and deploying machine learning models efficiently.

Location - HSRB Rollins Auditorium

1760 Haygood Dr NE, Atlanta, GA 30322

Continental Breakfast & Registration

Welcome to Emory

Keynote: Why AI Needs Feminism

Dr. Lauren Klein discusses the critical role of feminism in shaping the future of artificial intelligence.

Fairness in AI-based Software Development from Software Engineering Perspective

Explore strategies and insights on developing AI technologies that contribute positively to society.

Break

Deploying AI in Real World Setting

Gain insights from early evaluations of AI-Doc with Theo, understand real-world validation through the PRECISE trial with Siva Bhavani, and explore external validation of mammogram AI models with Hari. Join a panel Q&A session to delve deeper into deploying AI in practical applications Moderator : Dr. Enzo Ferrante

Breast Cancer Disparities: A Grand Challenge

Keynote by Lauren McCullough addressing the significant disparities in breast cancer, followed by a Q&A session to discuss challenges and solutions. Moderator : Dr. Janice Newsome

Lunch

Scaling collaborative learning - Update from the MIT Critical Data Network

Sadia Afreen

Keynote: AI Bias to AI by Us, for All of Us

Dr. Leo Celi explores the transition from AI bias to creating inclusive AI designed by and for everyone, followed by a Q&A session. Moderator: Dr. Po-Chi Kuo

Moving Beyond Race for Disparities Research

Explore strategies that move beyond race and sex groups for disparities research. Explore best practices for collecting skin tone data with Ian Wong, delve into genomics and ancestry considerations for health equity research with Manoj. Join a panel Q&A session to discuss these innovative approaches. Moderator: Dr. Idris Muhammad

Reception Location - Miller Ward Alumni House

There will be a shuttle bus transporting participants from the Symposium to the Reception on Friday, August 23rd. The bus will provide continuous transportation from HSRB to Miller Ward and Miller Ward to HSRB from 4:15 to 8:30. Participants returning to HSRB can take the C Shuttle to the Starvine parking deck to their cars. The first bus leaves HSRB at 4:15. The last bus returns to HSRB at 8:30

Opening Reception

Join us for food, entertainment, and networking as we kick off the datathon with an exciting opening ceremony.

Location - School of Medicine (SOM, Room 130)

100 Woodruff Circle, Atlanta, GA 30322

Datathon - Doors Open

Continental Breakfast & Registration

Welcome to Datathon

Dataset description : EMBED

Dataset Description : Emory CXR

Dataset description : Emory ICU dataset

Dataset Description : EnCoDE

Problem description and group overview

Group Photo

Team transitions to rooms

Datathon Begins

Mentoring Session

Datathon Continues

Lunch

Datathon Continues

Check-In with Groups

Group Exercise

Dinner

Location - School of Medicine (SOM, Room 130)

100 Woodruff Circle, Atlanta, GA 30322

Datathon - Day 2 - Doors Open

Continental Breakfast

Debrief

Datathon Continues

Mentoring Session

Final Slides & Code Submission

Lunch

Presentations

5 minutes allocated per group for their presentation.

Judging

Wrap Up & Results