In preparation, 2026
A unified plasma-proteomic model for multi-disease diagnosis and multi-horizon prognosis, aiming to enable earlier and more precise disease detection from a single blood draw.
Submitted to International Journal of Biostatistics · revise & resubmit
A residual-guided simulated annealing algorithm for building sparse, interpretable clinical risk scores, improving on existing integer risk-scoring methods.
Published in Lancet Oncology, 2024
An interpretable AI model to guide adjuvant imatinib treatment decisions after resection of localized gastrointestinal stromal tumors, validated in a large multi-institutional cohort.
Published in Lancet eClinicalMedicine, 2023
An interpretable AI model predicting recurrence risk after surgical resection of gastrointestinal stromal tumors, designed to support personalized post-operative surveillance and treatment planning.
Published in JAMA Surgery, 2022
An AI-driven analysis identifying optimal surgical margin widths in hepatectomy for colorectal cancer liver metastases to improve long-term oncologic outcomes.