NLP Crisis Context Mapper
Traditional text classification models evaluate crisis statements in a vacuum, focusing on isolated keywords rather than the temporal behavioral changes and context paths critical for clinical risk assessment.
The Solution
I developed an NLP engine that predicts crisis risk levels from user statements. Instead of just using TF-IDF vectors, I utilized a Neo4j Graph Database to map connections between emotional states, anxiety triggers, and temporal behavioral changes. A Scikit-Learn ensemble model categorizes the statements into clinical risk brackets.
- Wrote text preprocessing scripts utilizing TF-IDF, tokenizers, and sentiment analyzers in Python
- Designed a Neo4j graph schema to link nodes representing patient IDs, sentiment scores, and distress tokens
- Trained a Scikit-Learn ensemble model to categorize statements into clinical risk brackets
- Wrote Cypher queries to help evaluators trace chronological context pathways instead of isolated keywords
System Pipeline
Engineering Trade-offs
Why?I traded the simple setup of a relational database for the steeper learning curve of Neo4j. The graph architecture was necessary because tracking the relationship pathways between distress tokens proved far more predictive than their isolated frequencies.
Proven Impact
Enabled clinical evaluators to trace chronological context pathways, significantly enhancing pattern detection and reducing false-positive risk alerts by providing graph-backed contextual evidence for every prediction.