Zero-Shot Clinical Note Scoring

Built a zero-shot NLP pipeline using BioBERT-based Question Answering and Named Entity Recognition to score clinical patient notes against medical rubrics.

The prototype reached a 68.2% F1-score while reducing reliance on supervised data and improving explainability.