Projects
Spanning computational drug development, deep learning for clinical prediction, and HIPAA-compliant digital health platform engineering.
Heally Connect Digital Health Platform
Developing a digital health application for participant recruitment, survey administration, data collection, and study management to support the Heally Connect research study at Yale School of Public Health. Custom-built HIPAA-compliant platform integrating participant recruitment workflows, electronic survey administration, and longitudinal study management.
BiLSTM Sepsis Prediction Model
Built a BiLSTM deep learning model using MIMIC-IV real-world data (11,300+ ICU records) to predict sepsis onset. Created a blinded data review framework designing analytical pipelines to evaluate clinical safety signals — achieving 91% sensitivity for sepsis onset detection 4 hours in advance. End-to-end data pipeline in Python, BigQuery, and Streamlit validated against FDA SaMD principles.
ApoE4 Protein Misfolding Drug Development
Designing a novel small molecule therapeutic targeting ApoE protein misfolding in Alzheimer's disease, advancing from target identification through lead optimization for a CNS neurodegenerative indication.
- check_circle Conducting in silico ADMET modeling to assess blood-brain barrier/CNS permeability, hepatotoxicity risk, and cardiac toxicity profiling using predictive pharmacology frameworks.
- check_circle Building a bench-to-bedside development plan integrating target validation, safety pharmacology, and proof-of-concept study design for potential first-in-human evaluation.
