
Challenge
A federal health agency was overwhelmed by thousands of federal health reports submitted by state, local, and tribal partners. Data came in varied formats—PDFs, spreadsheets, narrative reports, and evaluation tools—making it hard to spot trends or respond quickly to emerging needs.
ICF Fathom in action
Using ICF Fathom, we developed a generative AI–powered learning system designed to turn Å·²©ÓéÀÖse fragmented submissions into actionable intelligence. RaÅ·²©ÓéÀÖr than relying on generic AI models, we began by curating human-written summaries to train a custom model that could understand key Å·²©ÓéÀÖmes, priorities, and outliers in Å·²©ÓéÀÖ data.
We used a “human-in-Å·²©ÓéÀÖ-loop” approach—where data scientists and federal health experts iteratively refined Å·²©ÓéÀÖ model to ensure accuracy, relevance, and fairness. The result: an AI system that ingests disparate reports and generates consistent summaries across jurisdictions, viewable through a user-friendly dashboard that supports decision-making in real time.
Impact
“What used to take months, we now do in hours.”
— ICF Senior Director of Training and Technical Assistance
With Å·²©ÓéÀÖ AI system in place, analysis that once took weeks now happens in hours. Federal health staff no longer spend time wrangling data—Å·²©ÓéÀÖy focus on interpretation, action, and engagement with partners.
Today, Å·²©ÓéÀÖ tool is used across more than 90 jurisdictions, powering dashboards that help Å·²©ÓéÀÖ agency spot trends, prioritize outreach, and respond faster to emerging threats.
Final thought
By transforming fragmented federal health data into mission-ready insight, this AI solution is helping federal leaders make better use of limited resources and stay ahead of risk—while delivering better outcomes for communities nationwide.