Program Results

Introduction to the event

Since joining in September 2023, I established a new research lab at the intersection of Machine Learning and Biomedical science. The group has rapidly grown to 14 trainees (4 undergrad students, 9 Master’s and 1 PhD student), enabling a interdisplinary research environement that balances methodological innovation with clinically motivated applications. Our current research program is organized around three pillars, including the study of cancer heterogeneity, manifold learning over time, and multiple‑instance learning with multi‑omics integration.

On the research front, we have delivered concrete advances across all pillars. In cancer heterogeneity, we built single‑cell atlases across multiple breast cancer datasets and paired these with spatial transcriptomics to relate histology to gene expression, including a two‑stage framework that combines contrastive pretraining with diffusion‑based decoding for image‑guided gene prediction. In temporal learning for healthcare, we developed multimodal approaches to predict adverse events from longitudinal EHR and continuous monitoring, addressing imputation, masking, and interpretability for real‑time use. In physiological sensing, we advanced cuffless blood‑pressure estimation from PPG/ECG using deep sequence models. In multi‑omics and MIL, we integrated time series and clinical text to identify patients at high risk of ICU admission (AUC > 0.90 on MIMIC‑IV) and demonstrated that combining skin sympathetic nerve activity with creatinine‑corrected urine biomarkers improves diagnosis of urge urinary incontinence (AUC 0.80 ± 0.07). We also initiated a metagenomics program linking microbial signatures to disease severity in pediatric CAP, now expanding to host genetics and COVID cohorts.

Teaching and curriculum development are core to my role. In the first two years, I offered eight courses (six graduate, two undergraduate), providing rigorous coverage of modern ML alongside domain‑specific case studies and reproducible practices.

To broaden perspectives and accelerate student growth, I have actively cultivated international visibility. I invited well‑established U.S. researchers to give guest lectures and seminars, exposing students to cutting‑edge methods and career pathways. Collectively, these efforts have built a high‑momentum, collaborative environment that advances science, trains talent, and positions the lab for sustained impact in machine learning for biomedicine.