Detailed Agenda

Sign in to the event

Use the Esri Events account associated with your registration to sign in.

Return to all sessionsReturn to all sessions
GI Science: GeoAI Benchmarks and Spatial Analysis of Fentanyl Contagion
Geographic Information Science experts present a new benchmark for LLMs running spatial analysis, and spatial analysis on fentanyl cases in the US.
Recommended Audience: data science, GeoAI

Presentations

  • GeoAnalystBench: Automating Spatial Analysis Workflow with Large Language Models
    Recent advances in Geospatial Artificial Intelligence (GeoAI) have been driven by generative AI and foundation models. While powerful geoprocessing tools are widely available in GIS, automating these workflows using AI agents remains a challenge, especially for non-expert users. This research explores the capabilities LLMs and AI agents such as ChatGPT, Claude, Gemini, Llama, and DeepSeek in automating GIS workflows and code generation. We introduce a benchmark of well-designed 50 real-world geoprocessing tasks carefully validated by GIS domain experts to evaluate these models.

    , University of Wisconsin-Madison

  • Catching Fentanyl: Spatial Contagion in the U.S. Opioid Epidemic, 1999–2024
    Fentanyl’s spread was analyzed using Global Moran’s I on county-level overdose deaths (1999–2024) from CDC WONDER. Prescription opioids showed weak spatial dependence driven by physician prescribing; heroin was urban then diffused rurally after pill-mill closures. Fentanyl began as localized outbreaks in Rust Belt cities (2006) and became regionally contagious after 2013. It did not become a national epidemic until spreading westward in 2019, when Moran’s I peaked, showing people can “catch fentanyl” much like an infectious contagious disease.

    , Geospatial Specialist, Northwestern Feinberg School of Medicine

Session Type: User Presentations
Session Level: All Attendees
Access Type: In-Person
Topic: GI Science, Public Health, Analytics and Data Science
Esri Products: ArcGIS Pro
Industry: Public Health, Higher Education
Keywords: AI, GeoAI benchmarks, fentanyl, spatial analysis