Biostatistics & AI for Biomedical Discovery (BIOSTAT-X)

Research Activity

BIOSTAT-X (Biostatistics & AI for Biomedical Discovery) is a methodological laboratory dedicated to turning complex biomedical data into reliable, clinically meaningful evidence. We provide expertise in classical biostatistics and study design, supporting observational studies and clinical trials with rigorous planning, analysis, and reporting.

Our team develops advanced trial methodology, including Bayesian adaptive and group-sequential designs, and principled use of historical information when appropriate.  We work on causal inference and efficient estimation strategies (e.g., propensity-score and doubly robust approaches) to strengthen real-world evidence.

A core activity is evidence synthesis through systematic reviews, meta-analysis and Bayesian network meta-analysis, including methods for handling missing data.

BIOSTAT-X builds and validates predictive models for risk stratification and decision support, with careful attention to calibration and clinical usability. Our AI/ML research spans interpretable machine learning to deep learning, applied to clinical, imaging and registry data for outcome prediction and monitoring. We also study how AI can enhance the design and conduct of clinical trials (e.g., participant selection, stratification, and data streams from digital devices). 

In biomedical discovery, we analyze high-dimensional omics data and perform integrative analyses across molecular layers (e.g., GWAS/meta-analysis and transcriptomics/RNA-seq).  We promote reproducible analytics pipelines and transparent reporting to facilitate translation from computational findings to robust biomedical knowledge.

We develop privacy-preserving approaches, including federated learning, to enable secure multi-centre collaboration.

The laboratory brings together academic researchers, PhD students, and Medical Statistics & Biometrics residents in a shared environment of research and training.