Christina X Ji
I finished my PhD in computer science at MIT this summer. My research in machine learning for healthcare was focused on 1) characterizing how providers make different treatment decisions using causal inference and statistics and 2) detecting changes over time in healthcare settings and updating ML models. I completed my MEng and BS in computer science at MIT in 2019. Resume
Research
I use machine learning, causal inference, and statistics to tackle clinical questions. Some projects include:
- Benchmarking different transfer learning approaches for tuning deep learning models for image classification from a sequence of distributions to the final time point
- Creating a statistical test to detect when a machine learning model needs to be updated due to distribution shift over time
- Analyzing real-world data to assess the causal effect of choice of doctor on the treatment decision
- Building a large language model to predict patient trajectories
- Evaluating off-policy reinforcement learning policies for sepsis treatment
Internships
I had the opportunity to explore different kinds of problems:
- Experimented with language models and diffusion-based graph neural networks to generate molecules for specific drug targets at Genesis Therapeutics in 2023
- Extracted data-driven insights on the causal effect of LinkedIn Learning features on engagement metrics and revenue at LinkedIn in 2021
- Built machine learning models to predict patient outcomes at Philips and IBM research in 2018
Teaching
I am passionate about teaching:
- Developed and taught a 4-week class on Introduction to Statistical Hypothesis Testing (6.S098) in January 2024
- Created course material and taught recitations as a teaching assistant for Introduction to Statistical Data Analysis (6.3720) in spring 2023
- Completed workshops on subject design, lesson planning, and inclusive teaching and earned a teaching certificate from MIT teaching and learning lab
I am honored to have received the following teaching and mentoring awards:
- Carlton E Tucker teaching award from MIT EECS in 2024
- Graduate student extraordinary teaching and mentoring award from MIT School of Engineering in 2024
- Featured associate advisor for first-year undergraduates in May 2019
Community service
I also care about building a welcoming community. To help incoming students find their place at MIT:
- Organized visit days and orientation for MIT EECS PhD program from 2020 to 2022
- Mentored under-represented students on PhD applications from 2020 to 2023 through the MIT EECS graduate application assistance program
- Advised first-year undergraduates from 2016 to 2020
- Led undergraduate orientation groups from 2016 to 2018
Publications
Characterizing variation in healthcare across time and providers using machine learning.
PhD thesis. 2024.
Assessing variation in first-line type 2 diabetes treatment across eGFR levels and providers.
Christina X Ji, Saul Blecker, Michael Oberst, Ming-Chieh Shih, Leora Horwitz, and David Sontag.
Under review. 2024.
[paper][code]
Seq-to-final: a benchmark for tuning from sequential distributions to a final time point.
Christina X Ji, Ahmed M Alaa, and David Sontag.
Under review. 2024.
[paper] [code]
Large-scale study of temporal shift in health insurance claims.
Christina X Ji, Ahmed M Alaa, and David Sontag.
Oral spotlight at Conference on Health, Inference, and Learning (CHIL) 2023.
[paper] [poster] [video] [code]
Finding regions of heterogeneity in decision-making via expected conditional covariance.
Justin Lim*, Christina X Ji*, Michael Oberst*, Saul Blecker, Leora Horwitz, and David Sontag. *equal contribution
Neural information processing systems (NeurIPS) 2021.
[paper] [poster] [video] [code]
Trajectory inspection: a method for iterative clinician-driven design of reinforcement learning studies.
Christina X Ji*, Michael Oberst*, Sanjat Kanjilal, and David Sontag. *equal contribution
American medical informatics association (AMIA) 2021 virtual informatics summit.
[paper] [video] [code]
Modeling progression of Parkinson's disease.
MEng thesis. 2019.
[thesis] [code]