Team Leader
Kathan Vyas
Texas A&M University
Computer Science
kathan@tamu.edu
Project Type
Research
Who Can Join
Masters Students, Undergraduate Students
Project Description
This project aims to revolutionize pediatric cardiovascular care by building interpretable and explainable AI dashboards for real-time clinical decision-making. Focused on infants with single ventricle congenital heart disease (SV-CHD) and at risk of shunt occlusion, we are developing two intelligent dashboards:
Chapter 1: A static risk prediction system using interpretable ML (e.g., Logistic Regression, XGBoost with SHAP) to identify high-risk patients and surface contributing features.
Chapter 2: A dynamic, longitudinal dashboard using time-series models (LSTM, GRU, hazard models) and LLMs to monitor evolving patient risk in real-time and provide actionable insights.
Both chapters integrate advanced visualization techniques using progressive disclosure and detail-on-demand to ensure clinicians access only the most relevant information—when and where they need it. We seek to close gaps in real-time usability, explainability, and fairness in AI for pediatric care.
Team Needs
I am assembling a cross-functional team of passionate individuals to help complete the analysis and dashboard development for Chapters 1 and 2. We’re looking for:
# Technical Skills
Data Analysts: Experience in Python (Pandas, scikit-learn, XGBoost), statistical modeling, and logistic regression.
ML Engineers: Knowledge of time-series modeling (LSTM, GRU, survival analysis), model interpretation (e.g., SHAP, LIME).
Frontend Developers / Visualization Experts: Familiarity with dashboard frameworks (e.g., Dash, Streamlit, React.js), and UX principles, especially progressive disclosure and detail-on-demand.
# Design & Research Skills
UX Researchers / Human Factors Designers: Understanding of cognitive ergonomics and clinician workflow.
Clinician Collaborators / Biomedical Experts: To validate medical relevance and usability of the dashboards.
# Bonus Perspectives
Prior experience in pediatric cardiology or critical care settings. (not at all necessary)
Passion for responsible AI, fairness, and healthcare equity.
Experience or interest in working with Large Language Models (LLMs) for medical reasoning.
Special Opportunities
– Real-world clinical impact: Help design tools that could save lives by empowering doctors with interpretable, timely, and trustworthy insights.
– Publication Opportunities: Co-authorship on high-impact papers focused on dashboard design, predictive modeling, and user studies.
– Mentorship & Learning: Work with a PhD researcher specializing in AI for healthcare and gain exposure to cutting-edge research in visualization, interpretability, and clinical decision support.
– Portfolio & Resume Booster: Gain hands-on experience in building real, deployable systems that integrate AI with user-centered design.