Predictive Model for Patient Stratification
Problem: A mid-size pharma company needed to identify which patients would respond best to a new oncology drug based on genomic data.
Approach: Developed a random forest model using RNA-seq and clinical data. Implemented rigorous cross-validation and feature importance analysis.
Outcome: Delivered a validated model with 85% predictive accuracy, enabling the design of a more targeted and cost-effective Phase II trial.