The relevance of spatial thinking and geographic techniques for the medical domain extends far beyond
the traditional bounds of using GIS and spatial analysis to investigate public health phenomena in
geographic space. With some imagination, medical concepts and practices can be seen as
simultaneously existing in a multitude of different spaces. Once such an overarching spatial viewpoint is
in place, health data can be transformed into engaging artifacts that support medical research,
education, and practice. For example, medical records generated by intensive care units could be turned
into nuanced representations of patients' health status. The written notes of medical professionals –
from surgeons to nurses and radiologists – can be a rich source of insight, not only with respect to
individual patients, but in surveying and monitoring the medical ecosystem. Meanwhile, natural
language processing and machine learning can be used to organize biomedical research results into
coherent knowledge structures. The presentation will highlight three projects that exemplify a spatial
approach to health informatics. These involve diverse data sources, namely (1) a corpus of two million
biomedical research articles, (2) detailed clinical records for more than 10,000 ICU patients, and (3) a
multi-year registry of several thousand heart attack patients.