Tutorial Description

The integration of geography with artificial intelligence (AI) presents a promising avenue for advancing decision support systems in public health. Geography serves as a critical dimension in numerous datasets, particularly but not exclusively in public health and epidemiology, where spatial relationships play a pivotal role. Incorporation of spatial description into AI and machine learning pipelines is therefore necessary to harness the full potential of geospatial data.

By considering the spatial aspect of data, AI algorithms can yield more accurate and insightful results, leading to enhanced decision-making processes in public health interventions. Spatially-enabled algorithms such as spatial clustering, geographically weighted and land-use regression, and neural networks tailored to geographic data hold immense potential in analyzing and interpreting complex spatial patterns inherent in public health datasets.

This tutorial aims to provide attendees with a comprehensive understanding of the significance of geography in AI applications within the realms of epidemiology and public health. Through practical examples and theoretical discussions, participants will gain insights into the main concepts of spatially-enabled data collection and analysis, and grasp some of the methodologies and tools necessary to integrate geographic information effectively into AI models, thereby enabling more informed and targeted public health interventions.

Audience

This tutorial is targeted to all AIME participants that wish to get familiar with the integration of geography in AI, learn some fundamentals on GIS creation and use, and explore some methodologies useful to this scope. The tutorial is intended to be of interest for researchers, experts and practitioners that come from a variety of backgrounds, such as computer science, bioengineering, public health, and epidemiology. Prerequisites that the participants are expected to possess are a basic knowledge of Data Mining, Machine Learning and AI algorithms. 

The tutorial will take place during the dedicated session at the AIME 2024 Conference in Salt Lake City, USA (https://aime24.aimedicine.info/), and will consist in a half-day session of lectures and discussions structured as follows:

• Introduction, background and relevance:

  • A little history: GPS tracking and spatial analysis

  • Fundamentals of exposomics

  • Spatial Enablement and its importance in public health and epidemiology

• Geographical Information Systems: creation and analysis

• Integration of geography in AI, overview and practical examples:

  • Spatial Clustering

  • Geographically Weighted Regression 

  • Land-use regression

  • Topological Data Analysis for spatial longitudinal data

  • Machine Learning and Deep Learning methods for spatially-enabled data

  • Agent-based Models and simulation tools

• Practical examples of applications in past and present academic projects

• Open challenges and opportunities