PharmaPulse

Inspiration Pharmaceutical companies operate in an environment where public perception can shift quickly due to safety concerns, regulatory discussions, or media narratives. Negative sentiment around a drug can escalate before internal teams even notice. We were inspired to build PharmaPulse as an early-warning intelligence tool that transforms scattered public news into actionable insights, helping pharma teams proactively monitor reputation and risk instead of reacting after escalation. What it does PharmaPulse is an AI-powered drug reputation monitor that analyzes public news related to pharmaceuticals and converts unstructured text into decision-ready intelligence. The system: Collects drug-related news articles Performs transformer-based sentiment analysis Groups discussions into thematic clusters Calculates a perception risk level Presents insights in an interactive dashboard Instead of simply displaying articles, PharmaPulse highlights which narratives are driving positive or negative perception and generates alerts that support faster communication and safety monitoring decisions. Challenges we ran into One of the biggest challenges was dealing with noisy public data. News feeds include irrelevant or mixed-context content, so filtering and preprocessing were critical to maintaining meaningful insights. Another challenge was translating raw AI outputs into human-readable decision signals. We wanted the system to go beyond charts and tables and actually communicate risk in a way that would make sense to non-technical stakeholders. We also had to balance ambition with hackathon constraints, ensuring the pipeline remained stable while integrating NLP models and clustering techniques. Accomplishments that we're proud of We successfully built an end-to-end AI pipeline that: Fetches live pharma-related data Applies transformer-based NLP models Identifies emerging discussion themes Generates actionable risk indicators Most importantly, we transformed analytics into decision-oriented insights rather than just visual outputs. PharmaPulse demonstrates how AI can support real-world pharmacovigilance and communication monitoring workflows in a lightweight and accessible way. What we learned Through this project, we learned how to design and deploy an applied NLP pipeline, integrate machine learning models into a web interface, and structure insights for decision-making rather than raw analysis. We gained hands-on experience balancing model performance, usability, and system reliability within hackathon time constraints. We also learned that presenting insights clearly is just as important as building the underlying intelligence system. What's next for PharmaPulse Future versions of PharmaPulse can expand into a full-scale pharmaceutical intelligence platform by integrating additional data streams such as: Clinical and medical literature databases Regulatory announcements and safety alerts Healthcare professional discussions Social sentiment tracking We also plan to enhance explainability, automated alerting, and predictive trend monitoring so that pharma teams can anticipate reputation risks rather than simply reacting to them.