Research
My research applies artificial intelligence and computational modeling to translational oncology, grounded in a decade of work with patient-derived cancer organoids and cell lines. Across five overlapping interests, the goal is the same: to turn molecular and pharmacological measurements on real human cancers into predictions that can guide therapeutic decisions.
Interests
1. Hit-to-Lead Optimization
Identifying promising chemical and biological agents and refining them toward clinically viable candidates. My recent work in this direction includes the enhancement of cisplatin efficacy in colorectal cancer organoids through pressurized intraperitoneal aerosol chemotherapy (Sci Rep, 2025), the antitumor screening of Bifidobacterium cell-free supernatants in colorectal organoids (Sci Rep, 2025), and the dissection of doxorubicin resistance mechanisms across 18 sarcoma cell lines (J Transl Med, 2024).
2. Computational Drug Efficacy Screening
Scaling drug response measurements across heterogeneous patient-derived models and modeling the response surface computationally. Multifocal organoid capture from individual colon cancers has shown that distinct lesions of the same tumor can carry distinct druggable vulnerabilities (Adv Sci, 2022; NPJ Genom Med, 2022; Biomed Pharmacother, 2022), and we have extended efficacy screening to non-standard conditions including simulated microgravity (Sci Rep, 2024).
3. Prediction of Optimal Treatment
Building models that map molecular features of a patient's tumor to the therapy most likely to benefit them. Recent efforts include clustering 26 pancreatic ductal adenocarcinoma cell lines into two distinct drug-sensitivity groups (Cancer Cell Int, 2025), characterizing drug response variability across 24 breast cancer cell lines and 3 organoids derived from malignant pleural effusions (Breast Cancer Res, 2025), and using organoid biobanks to predict metastatic potential in resectable pancreatic cancer (Cell Oncol, 2024).
4. AI-based Multi-omics
Integrating whole-exome sequencing, transcriptomics, and DNA methylation with machine-learning methods to extract clinically actionable signals. Examples from my work include integrative multi-omics analysis of matched primary and liver-metastatic colorectal cancer organoids to identify putative molecular targets (Transl Oncol, 2025), whole-genome bisulfite sequencing for stage- and subtype-specific methylation signatures in pancreatic cancer (iScience, 2024), and multiomic analysis of lung cancer pleural effusions revealing distinct druggable molecular types (Sci Rep, 2022).
5. Disease Modeling
The empirical foundation underneath the work above: systematic establishment, characterization, and biobanking of patient-derived organoids and cell lines that faithfully recapitulate the genomic architecture of the original tumor. Resources I have helped establish include 36 pancreatic cancer organoids (Cell Oncol, 2024), 24 breast cancer cell lines and 3 organoids (Breast Cancer Res, 2025), 18 sarcoma cell lines (J Transl Med, 2024), a glioblastoma line panel (Sci Data, 2023), 18 colorectal cancer cell lines (Sci Rep, 2020), and EUS-FNA-derived pancreatic cancer organoids (Gut Liver, 2022).
Collaboration
I welcome collaborations with clinicians, wet-lab investigators, and computational researchers interested in patient-derived models, AI-based multi-omics, or computational pharmacology. Please see the contact page.