D2K Capstone Projects
Spring 2024

In the D2K Capstone program, interdisciplinary teams of students (advanced undergrads, professional master's students, and Ph.D.) work on a semester-long real-world project sponsored by our D2K Affiliate Members.


Spring 2024 Projects


BCM Cardio

BCM Cardio


BCM Cardio | Machine Learning Approaches for Diagnosing Diastolic Dysfunction in Pediatric Patients

Currently, there is no recognized consensus approach to diagnose pediatric diastolic dysfunction (DD), resulting in unnoticed advancement towards diastolic heart failure. Our project hopes to enhance early detection and treatment by creating a machine learning model to classify pediatric DD. This model will use echocardiogram data gathered from pediatric patients treated at the Texas Children's Hospital.

Student Team Members

  • Santiago Aparicio
  • Yourong Bao
  • Leo Li
  • Kaitlyn Trushenski
  • Zhuowen Ye
  • Jeffrey Zhong
  • Jianguang Zhuang

Sponsor and Mentors

  • Sponsor: Dr. Sebastian Acosta, Dr. Minh Nguyen
  • D2K Fellow, PhD Mentor: Peixuan Jin
  • Faculty Mentor: Dr. Arko Barman

Watch 1-minute highlight video


BCM SIDS

BCM SIDS


Team Breath of Life | Utilizing Machine Learning to Identify Cardio-Respiratory Signatures that Predict Sudden Infant Death Syndrome (SIDS) in mouse models

The project aims to leverage machine learning to identify cardio-respiratory signatures predictive of adverse outcomes, specifically death, in mouse models of Sudden Infant Death Syndrome (SIDS). It involves developing new feature engineering techniques for analyzing cardio-respiratory waveforms, discovering patterns associated with disease outcomes, and refining machine learning models for real-time prediction. The data comprises several hundred experiments on control and distinct SIDS mouse models analyzed using custom software for waveform analysis. The project aims to apply successful algorithms to human neonate cardio-respiratory monitoring, contributing to understanding and preventing SIDS.

Student Team Members

  • Michael Popa
  • Suchir Misra
  • Amanda Hogan
  • Carlos Gonzalez Rivera
  • Bo Sung Kim
  • Surya Sunil

Sponsor and Mentors

  • Sponsor: Dr. Russell Ray, Dr. Chris Ward, Dr. Andersen Chang, Mahmoud Almadi, Baylor College of Medicine
  • D2K Fellow, PhD Mentor: Arda Bayer
  • Faculty Mentor: Dr. Arko Barman

Watch 1-minute highlight video


Climate Change

Climate Champions


Climate Champions | Long Term Weather Forecasting

Global warming presents a pressing challenge with far-reaching consequences for our planet's ecosystems and human societies. Our goal is to develop models with training data across a variety of cities in the US to predict two months of data six years from our training data showing the two-week predicted temperature averages starting from each day.

Student Team Members

  • Ibrahim Abdulrahmon
  • Jason Shi
  • Lindsey Jacobs

Sponsor and Mentors

  • D2K Fellow, PhD Mentor: Jasper Liao
  • Faculty Mentor: Xinjie Lan

Watch 1-minute highlight video


Course Evaluation

Course Evaluation


Team Course Evaluation | Exploring Sentiments of College Course Evaluation Comments: A Combination of Sentiment and Numerical Analysis of Rice University’s Course Evaluation Data from 2007-2023

Rice’s course evaluation results are crucial resources for both students in selecting courses and faculty in improving course quality. However, it is challenging to look over decades of evaluation data to find relevant information. Therefore, we apply machine learning techniques to extract sentiments and key insights from student feedback, saving time in navigating through evaluation results.

Student Team Members

  • Huijun Mao
  • Jacob Flynn
  • Longping Zhang
  • Rongyin He

Sponsor and Mentors

  • D2K Fellow, PhD Mentor: Janmajay Singh
  • Faculty Mentor: Dr. Su Chen

Watch 1-minute highlight video


Cybercare

Cybercare


CyberCare |

In the digital age, the widespread use of social media has greatly accelerated the dissemination of misinformation, presenting a significant challenge. To tackle misinformation regarding traffic on social media platforms like Twitter (now X), we have constructed a comprehensive dataset that integrates traffic-related tweets with traffic sensor data and incident reports. Additionally, we have developed a pipeline for detecting traffic-related misinformation on X, enhancing our ability to identify and address inaccuracies effectively.

Student Team Members

  • Sanjay Rajasekha
  • Frank Ran
  • Yifan Wu
  • Ningzhi Xu
  • Anthony Yan
  • Bryant Cassady

Sponsor and Mentors

  • Sponsor: Dr. Arlei Lopes da Silva
  • D2K Fellow, PhD Mentor: Yuxin Tang
  • Faculty Mentor: Dr. Arko Barman

Watch 1-minute highlight video


EMS

Team REMS


Team REMS | Improving REMS Service Plans Through Predictive Modeling

Rice Emergency Medical Services (REMS) serves the Rice community and provides them with accessible medical care. Since its establishment in 1996, REMS has seen significant increases in student enrollment, call volume (the number of calls), and its number of staff. Due to an expected addition of around 700 students to the student body in the next 3-5 years, Rice plans to build two new residential colleges to accommodate demand. Our project would like to predict the expected demand for REMS services for the next 3-5 years to help them plan for this increase.

Student Team Members

  • Zoe Katz
  • Kevin Cai
  • Yunli Su
  • Shenyuan Wu
  • Bill Huang
  • Chelsea Zhao

Sponsor and Mentors

  • D2K Fellow, PhD Mentor: Mehdi Zafari
  • Faculty Mentor: Xinjie Lan
  • Sponsor: Lisa Basgall

Watch 1-minute highlight video


Exam Scheduling

Exam Scheduling


Exam Scheduling Optimizers | Enhanced Strategies for Exam Scheduling Optimization: Integrating Mixed Integer Programming and Genetic Algorithms

Exam scheduling refers to the process of coordinating and structuring exams under specific constraints for students within an educational institution. Our team aims to implement a range of optimization techniques, including Mixed Integer Programming, Genetic Algorithms, and a novel combined algorithm to address the challenges posed by the exam scheduling problem.

Student Team Members

  • Kathryn Jarjoura
  • Teddy Gilman
  • Charlotte Cambor
  • Ryan Wang
  • Arielle Sanford

Sponsor and Mentors

  • D2K Fellow, PhD Mentor: Yihua Xu
  • Faculty Mentor: Xinjie Lan

Watch 1-minute highlight video


FE&P

Energenius Solutions


Energenius Solutions | Normative Energy Demand Modeling

We are collaborating with Rice University’s Facilities Engineering and Planning Department (FE&P), tasked with overseeing campus energy procurement, metering, and analysis. Our objective is to develop energy consumption models utilizing available utility sub-meter and weather data. These models aim to pinpoint energy waste at the building level, facilitating early detection of mechanical issues for prompt repairs.

Student Team Members

  • Jeff Brover
  • Tyler Braito
  • Anya Hansen
  • Patricia Hashimoto
  • Lucas Moreira Bogado
  • Natalia Mendiola

Sponsor and Mentors

  • Sponsor: Terie McClintock, Johnny Pickle, Joseph McGrath, Keaton Kinstley
  • D2K Fellow, PhD Mentor: Maryam Khalid
  • Faculty Mentor: Su Chen

Watch 1-minute highlight video


Paper Recommender System

Paper Pilot


Paper Pilot | Context-Aware Research Paper Recommendation System

Scientific literature is a cornerstone for all researchers, providing the foundation upon which they build their investigations. As scientific literature expands, the task of retrieving related and relevant literature becomes increasingly difficult, highlighting the need for efficient tools that can generate literature recommendations. Our project aims to relieve this by implementing a paper recommendation system based on text contextual surrounding citations of interest. In achieving this, we build a model powered by graphical neural networks to deliver research paper recommendations with low latency and take research efficiency to new heights.

Student Team Members

  • Sharath Giri
  • Judy Fang
  • Jerry Jiang
  • Jacky Jiang
  • James Murphy
  • Andres Villada

Sponsor and Mentors

  • D2K Fellow, PhD Mentor: Joao Pedro Mattos
  • Faculty Mentor: Arko Barman

Watch 1-minute highlight video


JobPosting Analysis

VerityVanguard


VerityVanguard | A Machine Learning Approach to Detecting Online Fraudulent Job Postings

In the digital era, job posting fraud presents a significant challenge that impacts the job market and online recruitment practices, leading to severe outcomes such as privacy breaches, financial losses, and tarnished organizational reputations. Aiming to protect thousands of millions of users’ and jobseekers’ privacy and provided with the relevant data regarding a job posting, our program develops an algorithm that detects whether the job listing is legitimate or fraudulent in nature.

Student Team Members

  • Henry Cui
  • Charlie Liu
  • Simon Wang
  • Cindy Xin

Sponsor and Mentors

  • D2K Fellow, PhD Mentor: Jiaao Zhang
  • Faculty Mentor: Xinjie Lan

Watch 1-minute highlight video


TCH LOS

TCH LOS Angeles


TCH LOS Angeles | Forecasting Length of Stay for Pediatric Critical Care Patients with Heart Disease: A Data-Driven Approach at Texas Children's Hospital

Nestled in the world’s largest medical center, Texas Children’s Hospital is one of the most influential pediatric hospitals in the U.S. One of its wings, the Cardiac Intensive Care Unit, specializes in treating children with heart disease and recovering from heart surgery. With highly sought after services, the CICU needs to effectively manage its capacity to serve as many patients as possible. Our machine learning model will estimate the length of stay for patients under 30 days of age admitted to the TCH CICU.

Student Team Members

  • Eli Ginsburg
  • Prayag Gordy
  • Derek Fu
  • Zev Lee
  • Kelly Zeng
  • Apple Li

Sponsor and Mentors

  • Sponsor: Di Miao, Christian Jenson
  • D2K Fellow, PhD Mentor: Nhi Le
  • Faculty Mentor: Xinjie Lan

Watch 1-minute highlight video


TXDot

BidOpt

BidOpt | Developing a quantitative bidding strategy for TxDOT fixed-price auction contracts using distribution regression.

Texas Department of Transportation (TxDOT) auctions ~$11B worth of statewide construction and maintenance projects each year. For each project, a number of firms submit sealed bids and the lowest bidder is awarded the contract and compensated with their bid amount. We propose a novel bidding strategy which uses historical bidding data and state-of-the-art distribution regression techniques to predict a project bid that optimizes a firm's expected profit.

Student Team Members

  • Benjamin Meisburger
  • Virginia Baskin
  • Daniel Cufino
  • Clay Kindiger
  • Alison Qui

Sponsor and Mentors

  • Sponsor: Hunter Follett
  • D2K Fellow, PhD Mentor: Jason Yang
  • Faculty Mentor: Dr. Su Chen

Watch 1-minute highlight video