News

New preprint on medRxiv
Our paper
Longitudinal AI Models Predict Cancer Adverse Events, led by Chenlian Fu, is now on medRxiv. SPARC is a transformer-based foundation model trained on routinely collected longitudinal clinical measurements from 224,047 patients with cancer; it predicts treatment-related adverse events and generalizes to external cancer centers.
Best Oral Presentation at ISMB 2026!
Our work
Routine Clinical Time Series Predict Cancer Adverse Events, led by Chenlian Fu with Dr. Justin Jee and Dr. Quaid Morris, received a Best Oral Presentation award at the 34th Conference on Intelligent Systems for Molecular Biology (ISMB 2026) in Washington, DC. I also presented our abstract Large language models for tumor genomic interpretation there (now on bioRxiv as Generalist large language models complement tailor-made predictors for tumor genomics interpretation).
New preprint on bioRxiv
My first-authored PhD paper,
Generalist large language models complement tailor-made predictors for tumor genomics interpretation, is now on bioRxiv. We evaluate general-purpose LLMs on three real-world tumor genomic interpretation tasks and show that they match or complement specialized predictors, with the largest gains on data from new assays and institutions.
Journal publication in JMIR Formative Research
Our paper
Benefit of the N-of-1 Approach Versus Aggregate Analysis in Tracking Individual Trajectories During Pregnancy, co-first-authored with Tina Behrouzi, was published in JMIR Formative Research. Using Oura Ring and survey data from pregnant participants in the BUMP study, we show that population-level trends often fail to reflect individual pregnancy trajectories.
Presenting at AACR Annual Meeting 2026 in San Diego
I presented our work
Large language models for tumor genomic interpretation at the American Association for Cancer Research (AACR) Annual Meeting 2026. Benchmarking generalist LLMs against tailor-made predictors on 100,000+ tumor samples from 19 institutions, we find that LLM–specialist ensembles substantially improve cross-institution accuracy. I also co-authored the abstract Allostatic load connects tumor genomics, disease trajectory, and pre-diagnosis wearable activity, led by Christopher Fong.
Paper at ML4H 2024 in Vancouver!
Our paper
Towards a personalized pregnancy experience: Forecasting symptoms using graph neural networks and digital health technologies, co-first-authored with Rui Zhu from my time in the Goldenberg Lab, was accepted to the Machine Learning for Health (ML4H) 2024 Symposium proceedings track and presented in Vancouver. The proceedings appear in PMLR (volume 259).
Clinical abstract at MLHC 2024
Our collaborative work with McMaster University,
Evaluating stimulus-based heart rate variability assessment in chronic constipation using a machine learning approach, was presented in the clinical abstract track of the Machine Learning for Healthcare (MLHC) 2024 conference in Toronto.