BIG DATA COMES TO PREDICTIONS OF PATIENT OUTCOMES. This article on the Quartz website by Dave Gershgorn describes how “Google is using 46 billion data points to predict the medical outcomes of hospital patients.” The possible outcomes for hospital admissions that are predicted include death, discharge, and readmission.
The research paper that announced this initiative had 34 co-authors. Interestingly, it was not peer reviewed.
The paper claims that it can predict death 24 to 48 hours quicker than existing methods.
Google used data from over 200,000 adults.
As with the Happyornot button, the paper doesn’t seem to make any attempt to sort out causal relationships among variables.