The Hospital as a Sorting Machine


Large hospitals sort patients into wards or clinics based on a number of factors. The information of a patient’s location is potentially useful in a number of medical contexts. By demonstrating and studying the sorting that occurs in a hospital, we show that each ward has characteristic signatures emergent from the blood test results associated with patients in those wards. We further demonstrate the utility of this information to fundamental research by evaluating the distribution of serum alanine aminotransferase, a liver enzyme hypothesised to be linked to heart disease. We demonstrate the creation of classifiers to determine to which ward a patient is assigned with greater than 60 % accuracy within the first three prioritised predictions across 28 wards. This study is extended to predict patients’ movements from intake wards through to destination wards as a proof-of-concept into patient routing. We propose that this study provides a basis for leveraging metadata in a number of fields, including: fundamental research, laboratory medicine quality assurance, hospital and admissions management, and patient triage.

Informatics in Medicine Unlocked
Benjamin Rosman
Benjamin Rosman
Lab Director

I am a Professor in the School of Computer Science and Applied Mathematics at the University of the Witwatersrand in Johannesburg. I work in robotics, artificial intelligence, decision theory and machine learning.